Sfp1 shuttles in a transcription-dependent manner and localizes to P-bodies.

(A) Shuttling assay. The assay used the temperature-sensitive (ts) nup49-313 mutant and was performed as reported (Lee et al., 1996; Selitrennik et al., 2006). Wild type (WT, yMS119), nup49-313(ts) (yMS1) and nup49-313(ts) rpb1-1(ts) cells expressing GFP-Sfp1 were allowed to proliferate under optimal conditions at 24°C. Cycloheximide (CHX) (50 μg/ml) was then added, and the cultures were shifted to 37°C (to inhibit Nup49-313 and Rpb1-1). The proportion of cells expressing cytoplasmic GFP-Sfp1 was plotted as a function of time (N>200). Error bars represent the standard deviation of three replicates. (B) GFP-Sfp1 is co-localized with P bodies markers. Cells expressing the indicated fluorescent proteins were allowed to proliferate till mid-logarithmic phase, followed by 24 h starvation in medium lacking glucose and amino acids. Live cells were inspected under fluorescent microscope. White arrows mark P-bodies. (C) The number of GFP-Sfp1 containing foci per cell decrease in response to cycloheximide (CHX) treatment. CHX (50 μg/ml) was added to exponentially proliferating cultures for the indicated time. Cells were then shifted to starvation medium as in B; Average of 2 replicates is shown. Student’s t-test between time 0 and the indicated time points was performed; ** represents p<0.001 (N≥240).

Sfp1 binds a group of mRNAs around GCTGCT motif.

(A) Metagene profile of CRAC analysis in two wild type replicates. RPKMs plot around the average metagene region of all yeast genes. The 5’UTR and 3’UTR are shown in real scale in base pairs (bp) whereas the transcribed region is shown as percentage scale to normalize different gene lengths. (B) Heatmap representation of CRAC reads around the polyadenylation site (pA) for the top 600 genes with the highest number of CRAC reads. The genes with high CRAC signal density upstream of the pA site are considered as CRAC+ (n=264), indicated on the right. The chosen cut-off was somewhat arbitrary; additional analyses shown in the subsequent figures indicate that this choice was biologically significant. (C) Average metagene analysis of two replicates of the CRAC+ signal in genes containing a GCTGCT motif in a region ±500 bp around the motif. CRAC reads were aligned by the center of the motif. (D) Sfp1 pulls down CRAC+ mRNAs. The extracts of isogenic cells, expressing the indicated tandem affinity purification tag (TAP), were subjected to tandem affinity purification (Puig et al., 2001), in the presence of RNase inhibitors. The RNA was extracted and was analyzed by Northern blot hybridization, using the probes indicated on the left. “+intron” denotes the position of intron-containing RPL30 RNA.

The mRNA-binding specificity of Sfp1 depends on the Rap1 binding site (RapBS) within the promoter.

(A) Constructs used in this study. The constructs were described previously (Bregman et al., 2011). To differentiate construct-encoded mRNAs from endogenous ones, we inserted an oligo(G)18 in the 3’ untranslated region. The constructs are identical except for the nature of their upstream activating sequence (UAS), located upstream of the ACT1 core promoter that includes the TATA box (designated ‘‘TATA’’). The nucleotide boundaries of the RPL30 sequences are depicted above the constructs, and those of the ACT1 sequences are depicted below the constructs. The numbering referes to the translation start codon. These constructs encode identical mRNA (Bregman et al., 2011). (B) Binding of Sfp1 to mRNA is dependent on Rap1-binding site (RapBS). Extracts of cells, expressing the indicated constructs, were subjected to RNA immunoprecipitation (RIP), using tandem affinity purification (TAP) of the indicated TAP-tagged proteins. RIP was followed by Northern blot hybridization using the probes indicated at the left. After the membrane was hybridized with oligo(C)18-containing probes (to detect RPL30pG mRNA; see Bregman et al., 2011), the membrane was hybridized with probes to detect endogenous RPL25 and RPL29 mRNAs. +Intron represents the intron-containing RPL30pG RNA (C) Most RiBi CRAC+ genes have Rap1 binding sites in their promoters. Only 25 RiBi genes (including RiBi-like) are defined as CRAC+. Upper panel: Venn diagram showing overlap between these genes and genes carrying promoters with RapBS. Lower panel: Venn diagram showing overlap between all RiBi (+ RiBi-like) genes excluding CRAC+ (“non-CRAC+”) and genes carrying promoters with RapBS. Data on Rap1 bound promoters were obtained from (Lieb et al., 2001). A hypergeometric test was applied to calculate the p-values, indicated underneath each diagram. Note that p-values were significant for both inclusion (upper diagram) and exclusion (lower diagram).

Sfp1 is required for efficient transcription of CRAC+ genes and stabilizes the deadenylation-dependent pathway of their mRNAs’ decay.

(A) Sfp1 deletion affects SR (synthesis rate), RA (mRNA Abundance) and HL (mRNA half-life) differently for different subsets of genes. CRAC+ (n=264), RiBi (including RiBi-like) CRAC+ (n=25), RiBi (+RiBi-like) excluding CRAC+ (“RiBi (non-CRAC+) (n=411), RPs (n=129). Statistical analyses were performed using the Wilcoxon test. Asterisks indicate significant results (* p-value < 0.05; ** p-value < 0.01; *** p-value < 0.001; **** p-value < 0.0001). Unless indicated otherwise, statistical comparisons were performed using the “Rest” group (all detectable genes exclusing CRAC+ ones) as a reference. In addition, RiBi CRAC+ and RiBi non-CRAC+ are compared against each other. (B) Sfp1 depletion rapidly affects SR and RA of CRAC+ genes. Scatterplot of changes in SR vs changes in RA at 20 min. (left) or 60 min. (right) after depleting Sfp1-degron by auxin. CRAC+ genes are highlighted in green. Spearman correlation values and the significance of the linear adjustment for the whole dataset are indicated inside the plot. Density curves are drawn on the margins of the plot to help evaluate the overlap between dots. (C) RapBS confers Sfp1-dependent mRNA decay pathway. Shown is the quantification of Northern blot hybridization results of mRNA decay assay (Methods), performed with WT or Δsfp1 cells that carried the indicated constructs (described in (Bregman et al., 2011) and shown at the top). The membrane was probed sequentially with an oligo(C)18-containing probe, to detect the construct-encoded mRNA, and with probes to detect endogenous mRNAs. mRNA levels were normalized to the Pol III transcript SCR1 mRNA (Methods). The band intensity at time 0, before transcription inhibition, was defined as 100% and the intensities at the other time points (min) were calculated relative to time 0. Error bars indicate the standard deviation the mean values of three independent replicates (for (G)18-containing mRNAs), or of 12 replicates (for endogenous mRNAs). (D) The deadenylation rate of CRAC+ mRNAs and the subsequent decay of the deadenylated RNAs is accelerated in sfp1Δ cells. Transcription was blocked as described in section C (Methods). RNA samples were analyzed using the polyacrylamide Northern technique (Sachs and Davis, 1989), using the probes indicated on the left, all are CRAC+. Half-lives were determined and the indicated ratios are depicted on the right. The asterisk (*) indicates the time point at which deadenylation is estimated to be complete.

Binding features of Sfp1 to chromatin.

(A) Sfp1 is present in the bodies of CRAC+ genes. Average metagene of Sfp1 and Rap1 ChIP-exo signal, obtained from (Reja et al., 2015), for CRAC+ genes (n=264) and CRAC-CONTROL genes (a subset of 264 non-CRAC+ genes randomly selected from the entire genome, but excluding RP and RiBi genes). See also the plot with alternative scaling in Fig. S4A). (B) Left panel: Positive correlation between Sfp1 binding to gene bodies and transcriptional activity. Scatterplot comparing Sfp1 binding to gene bodies (measured by ChIP-exo, Reja et al., 2015) versus the density of actively elongating RNA pol II (measured by BioGRO-seq; Begley et al., 2021) in CRAC+ genes (n=264). Spearman correlation is indicated at the bottom of the plot. Right panel: Correlation between Sfp1 binding to the bodies of all genes (n=4777) and transcription rate. Spearman correlation for all genes (including CRAC+) or for all genes exclusing CRAC+ genes are indicated at the bottom of the plot. (C) The Sfp1 ChIP-exo signal drops downstream of the GCTGCT motif. Comparison of average metagene profiles of CRAC (blue) and ChIP-exo (orange) signals for genes with a GCTGCT motif (n=163).

Sfp1 induces Pol II backtracking, preferentially in CRAC+ genes.

(A) Sfp1 differently affects Pol II occupancy (left) and on Pol II activity (right). Pol II levels were measured by Rpb3 ChIP and its activity - by genomic run-on (GRO), in WT (BY4741) and its isogenic sfp1Δ strain, growing exponentially in YPD. Anti Rpb3 (rabbit polyclonal) ChIP on chip experiments using Affymetrix® GeneChip S. Cerevisiae Tiling 1.0R custom arrays as described in Methods. For each gene, the average of signals corresponding to tiles covering 5’ and 3’ ends (250 bp) were calculated. Green dots represent the CRAC+ genes; grey dot – non-CRAC+. The tendency line, its equation, Pearson R and its p-value of the statistically significant deviation from the null hypothesis of no correlation are shown in grey, for the whole dataset, and in green, for the CRAC+ genes. (B) Sfp1 promotes Pol II backtracking of CRAC+ genes. Backtracking index (BI), defined as the ratio of Rpb3-ChIP to GRO signals, is shown for different gene sets, indicated below, comparing WT and sfp1Δ strains. In order to compare data obtained from different types of experiments the values were normalized by the median and standard deviation (z-score). The bars represent standard errors. Statistical significance of the differences between the averages of the indicated samples was estimated using a two-tailed Student’s t-test (* means p < 0.01). (C) mRNA HL and BI of CRAC+ mRNAs/genes are affected by Sfp1. Box and whisker plots showing the effect of Sfp1 on mRNA HL and Pol II BI. A comparison between CRAC+ (green) and all genes excluding CRAC+ ones (“non-CRAC+” grey) genes is shown. HL was calculated from the mRNA abundance (RA) and synthesis rates (SR), using the data shown in Fig. 4A. The statistical significance of the differences between the averages of the CRAC+ and non-CRAC+was estimated using a two-tailed Student’s t-test (*** means p < 0.0001). (D) mRNA HL and BI are correlated via Sfp1: correlation between Sfp1-dependence of BI and mRNA HL ratios. Data from C were represented in a scatter plot. Linear regression equations are shown for all (grey) and CRAC+ genes (green). Pearson correlation coefficient, r, and the p-value of the statistically significant deviation from the null hypothesis of no correlation (r = 0) are also indicated. All statistical correlations were determined using the ggpubr package in R.

Sfp1 alters Rpb4 stoichiometry/configuration within Pol II elongation complex and this alteration is linked to mRNA stabilization.

(A) The Rpb4 stoichiometry/configuration changes along the transcription units in Sfp1-mediated manner. Top left panel - the values of Rpb3 and Rpb4 were obtained from ChIP on chip experiments, either against Rpb3 or against Rpb4-Myc in LMY3.1 cells proliferated exponentially in YPD. Rpb4-ChIP/ Rpb3-ChIP ratios were calculated and averages for the indicated genes sets were obtained for positions ranging from -100 to +250 (relative to TSS) and from -250 to +100 (relative to pA sites). Average ratios were normalised to the TSS -100 position, in order to represent profiles of Rpb4-ChIP changes after Pol II recruitment to promoters. Top right panel –profiles of Rpb4-ChIP/Rpb3-ChIP ratios were obtained as in top right panel, but from sfp1Δ strain (LMY7.1). Bottom left panel - Rpb4- ChIP/Rpb3-ChIP profiles of CRAC+ genes: comparing WT (blue) and sfp1Δ (red) strains. Bottom right panel - Rpb4-ChIP/Rpb3-ChIP profiles of rest of the genes (all detectable genes excluding CRAC+ ones) : comparing WT (blue) and sfp1Δ (red) strains. (B) Correlation between the Rpb4-ChIP/Rpb3-ChIP ratios and Pol II BI in WT (left panel) or in sfp1Δ cells (right panel). For each gene, average of Rpb4-ChIP/Rpb3-ChIP values corresponding to positions from TSS to +250 and from -250 to pA sites were calculated. BI values taken from Fig. 6C. Linear regression equations are shown for all (grey) and CRAC+ genes (green). Pearson correlation coefficients, r, and the p-values of the statistically significant deviation from the null hypothesis of no correlation (r = 0), are also shown. All statistical correlations were determined using the ggpubr package in R. (C) Correlation between the Rpb4-ChIP/Rpb3-ChIP ratios and mRNA HL in WT or sfp1Δ cells. Values of Rpb4-ChIP/Rpb3-ChIP ratios ere determined as in B. mRNA HL was indirectly calculated from mRNA abundance and transcription rates taken from the data used in Fig. 4A and is shown in arbitrary units. CRAC+ genes are depicted in green. R was calculated as in B.

A model for Sfp1 functions in yeast.

(A) Two modes of action of Sfp1 in transcription: CRAC+ genes recruit Sfp1 to their promoters, whereas non-CRAC+ genes recruit Sfp1 from the nuclear space directly to Pol II. Upper panel represents CRAC+ genes that recruit Sfp1 to their promoters. We discovered that Sfp1 appears to accompany Pol II of CRAC+ genes, in a manner proportional to the number of transcriptionally active Pol II (Fig. 5A-B). Its binding to all Pol II molecules, including backtracked Pol II, is even more apparent (Fig. 5B). In addition, we found that CRAC+ genes are enriched with Rap1-binding sites (Fig. 3). We therefore propose that, following binding to Rap1-containing promoters, Sfp1 binds Pol II. Specifically, it binds to Rpb4 (and possibly other Pol II subunits) and accompanies it until imprinting. This interaction influences Pol II configuration (Fig. 7A-D) and increases the likelihood of Pol II to undergo backtracking (Fig. 6B). Lower panel represents non-CRAC+ genes that also interact with Sfp1 (Fig. 5A CONTROL, Fig. 5B). We propose that promoters of non-CRAC+ genes recruit Sfp1 poorly (relative to CRAC+ promoters), except for small group of promoters, e.g., of RiBi genes lacking RapBS. The dashed arrow represents this minor group. For the majority of these genes, the nuclear Sfp1 interacts directly with their elongating Pol II, as its interaction correlates with the extent of chromatin-bound Pol II (Fig. 5B). This weak interaction also changes Pol II configuration (Fig. 7A, “non-CRAC+”) and increases the propensity of Pol II to backtrack (Fig. 7B). However, it either does not result in imprinting or results in rare imprinting events that went undetected by our CRAC assay. (B) Sfp1 mediates cross-talk between mRNA synthesis and decay via imprinting. The backtracked configuration, induced by Sfp1 (A), is compatible with a movement of Sfp1 from Pol II to its transcripts (see text), which is enhanced in case the GCTGCT motif is localized near Sfp1 (Fig. 5C). Following co-transcriptional RNA binding, Sfp1 accompanies the mRNA to the cytoplasm and stabilizes the mRNAs. Following mRNA degradation, Sfp1 is imported back into the nucleus to initiate a new cycle. The model proposes that the specificity of Sfp1-RNA interaction is determined, in part, by the promoter (Fig. 3A-B). Nevertheless, promoter binding is necessary, but not sufficient for RNA binding. See text for more details.