Fig 1.
Sumoylated proteins associate stably with chromatin, including at many non-RPGs.
(A) Most sumoylated proteins are associated with chromatin. Fractionation experiments were carried out in normally growing, unmodified yeast and human HeLa cells, then whole cell extracts (WCE) and indicated fractions were analyzed by immunoblot (IB) with indicated antibodies, including antibodies for the yeast Smt3 peptide (“SUMO”), human SUMO1 or SUMO2/3 isoforms, and histone H3. (B) Independent duplicate SUMO ChIP-seq experiments were performed in yeast, and sample alignments are shown from Replicate 1, using the Integrative Genomics Viewer (IGV) genomic alignment tool, with reads from the SUMO IP and corresponding inputs over selected short segments of Chromosomes XII and IV. Values correspond to maximum data range (read numbers) for the view shown and blue bars along the bottom represent gene positions, including for two non-RPGs, CCW12 and LYS20, whose ORF orientations are indicated with arrows. See S9 Table for a list of yeast genes associated with the 603 SUMO peaks. (C) Pie chart showing fractions of the 603 SUMO ChIP-seq peak set associated with different gene types. Non-RPG refers to protein-coding genes that are not ribosomal protein genes (RPGs). See S2 Table for detailed description of peak classifications. (D) Distribution of read counts for the 603 SUMO ChIP-seq peaks, separated by gene type, then ranked by normalized read counts. Normalized read counts were determined using DiffBind tool with two independent ChIP-seq replicates. (E) Sample SUMO peak alignments from Replicate 1 are shown for two RPGs and four non-RPGs. Peak alignments for TBP and Rap1 are also shown, using published ChIP-seq and ChIP-exo datasets, respectively, for comparison with SUMO peak positions (NCBI GEO database accession numbers GSM2870615 and GSE93662, respectively). Values refer to maximum data range (read numbers) for the view shown, and unnormalized alignments were generated using IGV.
Fig 2.
Many highly transcribed non-RPGs have a promoter-associated SUMO peak.
(A) Comparison of SUMO peak-containing genes by transcription level. The 111 SUMO peaks (blue) associated with single non-RPGs were plotted, by normalized read count, over the ranked distribution of RNAPII densities (pink) of all protein-coding genes (PCGs), which are approximations of transcription levels and were determined by duplicate independent RNAPII ChIP-seq analyses performed in the same strain and conditions as the SUMO ChIP-seq. Inset shows 12 SUMO peaks of the set of 111 that are associated with genes that have no detected RNAPII density. See S11 Table for list of RNAPII densities by gene. (B) Box plot comparing RNAPII densities for all RNAPII-occupied genes and for the 111 non-RPGs with unique SUMO peaks. P-value of Student’s t-test analysis is shown. (C) Composite plots of RNAPII ChIP-seq profiles in WT or ubc9-6 cells at non-RPGs with the highest RNAPII densities (top 10%) that either contain or lack promoter-associated SUMO peaks, as indicated. Plots were generated using the ComputeMatrix tool (from deepTools 3.3.0). TSS, transcription start site; TES, transcription end site. (D) Composite plots of SUMO peaks associated with single non-RPGs in WT or ubc9-6 cells, shown relative to nearest TSS, generated using ComputeMatrix. (E) Unnormalized ChIP-seq peak alignments for SUMO and RNAPII in WT and ubc9-6 cells at four selected genes with normally high RNAPII densities (i.e. in WT cells). Values refer to maximum data range (read numbers) for the view shown per gene for both WT and corresponding ubc9-6 alignments. (F) Plot of RNAPII densities of all protein-coding genes in WT versus ubc9-6 cells, with densities at the 111 non-RPGs associated with unique SUMO peaks highlighted. See S11 Table for list of RNAPII densities in WT and ubc9-6 strains by gene.
Fig 3.
SUMO peaks at promoters of non-PRGs align specifically with GTF components.
(A) Comparison of SUMO and TBP composite plots at SUMO peak-containing RPGs (left) and non-RPGs (right), shown relative to the TSS position. SUMO plots were generated with data from Replicate 1 of our SUMO ChIP-seq analysis, whereas TBP data were obtained from NCBI GEO database (accession number GSM2870615). (B) Comparison of SUMO and RNAPII (subunit Rpb3) composite plots at SUMO peak-containing non-RPGs. Rpb3 data were obtained from the GEO database (accession number GSM3629815). (C) Composite plots for GTF subunits, histone H3, sequence-specific transcription factors (SSTFs), and Mediator subunits were compared with SUMO composite plots at SUMO peak-containing non-RPGs. All datasets, except for SUMO (as described in A), were obtained from the GEO database, with the following accession numbers: Sua7 (TFIIB subunit): GSM4319112; Tfa2 (TFIIE subunit): GSM4319116; Tfg2 (TFIIF subunit): GSM4319120; H3: GSM2561057; Sko1: GSM3335975; Msn4: GSM1859030; Med8 (Mediator head subunit): GSM3189528; Med14 (Mediator tail subunit): GSM3189529; and Med3 (Mediator tail subunit): GSM3189530. (D) Distribution of SUMO peak-containing non-RPGs by TFIIF occupancy level. The 111 SUMO peaks (blue) associated with single non-RPGs were plotted, by normalized read count, over the ranked distribution of TFIIF (subunit Tfg2) occupancy levels (ChIP-seq counts per million) over all Tfg2-containing non-RPGs. Tfg2 occupancy levels derived from GEO database (accession GSM4319120).
Fig 4.
The large subunit of TFIIF, Tfg1, is sumoylated at Lys 60/61.
(A) Yeast strains were generated that each express a 6xHA C-terminal epitope tag on a different GTF subunit, including Toa1 (TFIIA), Taf7 (TFIID), Taf8 (TFIID), Tfg1 (TFIIF), or Tfa1 (TFIIE). Cultures of the strains were used to prepare lysates, under non-denaturing conditions, that were then used in HA IP experiments, followed by SUMO and HA immunoblots (IBs). (B) Strains expressing HA-tagged forms of TFIIF subunits Tfg1, Tfg2, or Tfg3 (also known as TAF14), were used for HA-IP experiments followed by SUMO and HA immunoblot analysis. “No tag” refers to the parental strain (W303a) that expresses no HA-tagged proteins. (C) Strains expressing Tfg1-HA, Tfg2-HA, or Tfg2-9HA (with a 9xHA tag instead of the usual 6xHA tag) were used in an HA IP analysis performed with lysates generated under non-denaturing conditions, followed by HA and SUMO immunoblots. The pattern detected in the SUMO blot of the Tfg2 IPs likely corresponds to sumoylated Tfg1 which coIPs with both forms of Tfg2. In the Tfg1-HA IP lane, the sumoylated species migrate slower because sumoylated Tfg1-HA includes the 6xHA tag, whereas Tfg1 (and its sumoylated forms) is untagged in the Tfg2-6HA and Tfg2-9HA strains. (D) Lysates were prepared from Tfg1-HA and Tfg2-9HA strains, which were then either treated (+) or not treated (-) by boiling for 5 min and adjusting the NaCl concentration to 0.5 M to promote the disruption of protein complexes prior to IP. Lysates were then used for HA IP analysis followed by SUMO and HA immunoblots. The pattern seen in the SUMO blot of the Tfg2-9HA IP disappears in the treated sample, implying that these sumoylated species are indeed derived from coIPed, sumoylated Tfg1. Note that treatment appears to elevate Tfg1 sumoylation levels (compare first two lanes in SUMO blot), likely because it inactivates naturally occurring SUMO proteases present in the lysate. (E) A mutant strain was generated that expresses Tfg1-HA with Arg substitutions at Lys 60 and 61 (K60,61R). This strain, along the WT Tfg1-HA-expressing strain, were used in HA IP-immunoblot experiments, and SUMO and HA immunoblots are shown. To disrupt protein-protein interactions, NaCl concentration was increased to 0.5 M and lysates were then boiled for 5 min, then cooled on ice prior to IP. Inputs represent approximately 5% of the material used for IP. (F) Diagram of Tfg1 domain structure, based on [64]. The major SUMO acceptor site is indicated with an encircled S at Lys 60/61. (G) Sumoylated Tfg1 associates with chromatin. A strain harboring a point mutation (I615N) in ULP1 (mt), and an isogenic wild-type strain (WT), were engineered to express Tfg1-HA, then the strains were used for chromatin fractionation analysis. Whole-cell extract (WCE), soluble, and chromatin fractions were analyzed by HA, histone H3, and GAPDH immunoblots. Red arrowheads indicate the position of mono-sumoylated Tfg1 in HA and SUMO immunoblots throughout.
Fig 5.
Elevated Tfg1 sumoylation reduces its interaction with RNAPII.
(A) Tfg1-HA and Tfg1-K60,61R-HA strains, and strains expressing Tfg1-HA in the ulp1-mt (I615N) or ULP1 parental backgrounds, were used for co-IP experiments. HA IPs, and no antibody (No Ab) controls, were analyzed by HA and Rpb1 (8WG16 antibody) immunoblots. Relative levels of co-IPed Rpb1, normalized to Tfg1-HA IP levels, were determined by densitometry and average values from at least three experiments are shown. Student’s t-tests were performed and paired values that are statistically different (p value < 0.05) are indicated with an asterisk. (B) Using the same strains described in A, Rpb1 IP was performed, followed by HA and Rpb1 immunoblots. Quantification of levels of co-IPed Tfg1-HA, normalized to IPed Rpb1 levels, was performed as in A. (C) Strains expressing Tfg1-HA with an N-terminal fusion to the yeast SUMO peptide (SUMO-Tfg1) or to a mutant form of the SUMO peptide in which all Lys residues are replaced with Ala (mSUMO-Tfg1), were analyzed alongside strains expressing WT and K60,61R forms of Tfg1-HA in a HA IPs, followed by SUMO and HA immunoblots. (D) Strains expressing WT, K60,61R, SUMO-Tfg1, or mSUMO-Tfg1 forms of Tfg1-HA were used for Rpb1 IP followed by immunoblot analysis with HA and Rpb1 antibodies.
Fig 6.
Both reducing and constitutively increasing promoter-associated SUMO levels can reduce RNAPII gene occupancy.
(A) SUMO ChIP was performed in Tfg1-HA and Tfg1-K60,61R-HA strains followed by qPCR analysis of promoter regions of the indicated genes, including a non-RPG that lacks a significant SUMO peak, TDH1, two non-RPGs that contain a promoter-associated SUMO peak, PDC1 and PYK1 (also known as CDC19), and an RPG, RPS20. Quantification was performed by normalizing to the background ChIP signal for an untranscribed region of Chromosome V for which there is no detectable SUMO signal in the SUMO ChIP-seq analysis. Average and standard deviation of three replicates is shown, and Student’s t-tests showing significant difference (p value < 0.05) between paired samples are indicated with an asterisk. (B) RNAPII ChIP-seq was performed, with two independent replicates, in Tfg1-HA and Tfg1-K60,61R-HA strains, and RNAPII densities (log2 normalized read counts) were determined at each gene ORF and plotted. Differential binding analysis was performed, using the DiffBind tool, and genes showing significantly different levels of RNAPII (FDR < 0.05) in the two strains are indicated with red dots. See S12 Table for list of RNAPII densities by gene in Tfg1-HA and Tfg1-K60,61R-HA strains. (C) Sample RNAPII ChIP-seq alignments from Replicate 1 are shown for genes showing varying levels of differential RNAPII densities, including three non-RPGs that contain a prominent SUMO peak at their promoters. Values refer to maximum data range (read numbers) for the view shown, and unnormalized alignments were generated using IGV. (D) SUMO and RNAPII ChIP were performed in strains expressing Tfg1-HA or SUMO-Tfg1-HA, followed by qPCR analysis of promoter regions of the indicated genes. Average and standard deviation of three replicates is shown, and Student’s t-tests showing significant difference (p value < 0.05) between paired samples are indicated with an asterisk. (E) Position of Tfg1 Lys 60/61 within the structure of the yeast transcription initiation complex. Enlarged section at right shows the disordered 62-amino acid segment of Tfg1 (dashes between Arg 35 and Glu 97) within which lies Lys 60/61, and a proximal segment of Rpb2 including a loop and helix of the External 1 region. Tfg1 is represented by a pink ribbon, Rpb2 by dark orange, and Tfg2 by grey. Image generated with RCSB PDB (rcsb.org), with PDB ID: 5FZD [42].