HSV-1 Infection Induces a Downstream Shift of Promoter-Proximal Pausing for Host Genes

Figures

• 

  FIG 1. HSV-1 infection impacts promoter-proximal pausing of most host genes. (a) Scatterplots comparing pausing indices (PI) between mock and WT-F infection at 3 h p.i. The dashed line indicates equal PI values and solid lines a 2-fold change in PIs. Genes were divided into three groups according to changes in their PI: (i) increased PI in HSV-1 infection (fold change >1, 763 genes, red); (ii) slightly reduced PI in HSV-1 infection (fold change <1 but ≥0.5, 2,082 genes, blue); and (iii) strongly reduced PI in HSV-1 infection (fold change <0.5, 4,211 genes, green). (b) Read coverage around the TSS in PRO-seq data (sense strand only) for mock (dark green) and WT-F infection at 3 h p.i. (dark blue) for example genes with a reduction in PI upon HSV-1 infection. Read coverage was normalized to total number of mapped reads and averaged between replicates. The identified TSS used in the analysis is indicated by a short vertical line below each read coverage track. Gene annotation is indicated at the top. Boxes represent exons, lines represent introns, and direction is indicated by arrowheads. Genomic coordinates are shown at the bottom. Figures are not centered around the TSS, but a larger…

• 

  FIG 2. Distinct patterns of changes in promoter-proximal pausing upon HSV-1 infection. (a) Heatmap showing the result of the hierarchical clustering analysis of PRO-seq profiles in mock and WT-F infection. For clustering, PRO-seq profiles in sense direction for mock and WT-F infection were first concatenated and then divided by the maximum value in the concatenated profiles. This resulted in a value of 1 for the position of the highest peak in either mock or HSV-1 infection. Hierarchical clustering was performed according to Euclidean distances and Ward’s clustering criterion, and the cutoff on the hierarchical clustering dendrogram was selected to obtain 50 clusters (marked by colored rectangles between the dendrogram and heatmap). Clusters are numbered from top to bottom. (b to e) Metagene plots of PRO-seq profiles on the sense strand in mock (dark green) and WT-F infection at 3 h p.i. (dark blue) for example clusters 6, 4, 47, and 9 in panel a. See Materials and Methods and Fig. 1 legend for an explanation of metagene plots. The colored bands below the metagene curves in each panel indicate the significance of paired Wilcoxon tests comparing the normalized PRO-seq coverages of genes…

• 

  FIG 3. Downstream shift of Pol II pausing throughout the first 6 h of HSV-1 infection and impact of ICP22. Metagene plots of PRO-seq profiles on the sense strand for 1.5-, 3-, and 6-h p.i. WT-F infection (a and b), repair virus infection (c and d), and ΔICP22 infection (e and f) for example clusters in Fig. 2a. Metagene plots with significance analyses of pairwise comparisons between time points for WT-F infection for these and other example clusters can be found in Fig. S6. Metagene plots with significance analyses of pairwise comparisons between repair virus and ΔICP22 infection for the same time points for these and other example clusters can be found in Fig. S7.

• 

  FIG 4. Changes in antisense promoter-proximal pausing in HSV-1 infection. Metagene plots of PRO-seq profiles on antisense direction in mock (gold) and WT-F infection at 3 h p.i. (red) for example clusters resulting from the hierarchical clustering of genes according to antisense PRO-seq profiles in mock and WT-F infection. Here, clustering was performed as described in Fig. 2a legend but applied to concatenated antisense PRO-seq profiles in mock and WT-F infection. Thus, clusters shown here differ from the clusters shown in all other figures. (a) The most common pattern observed for almost all clusters with a broadening of the antisense PRO-seq peak at the TSS. (b to d) The only three clusters that exhibit different patterns with additional peaks originating or increasing in antisense direction during infection. See Materials and Methods and Fig. 1 and 2 legends for the explanation of metagene plots.

• 

  FIG 5. Delayed pausing is not an artifact of alternative de novo initiation or read-in transcription. (a to d) Metagene plots of cRNA-seq profiles on the sense strand in mock and WT-17 infection at 1, 2, 4, and 8 h p.i. (a and b) and dRNA-seq profiles on the sense strand in mock and 8-h p.i. WT-17 infection (c and d) with and without XRN1 treatment for example clusters 4 and 9, which show broadening of peaks or additional peaks originating or increasing in height in PRO-seq data during WT-F infection. For metagene plots of PRO-seq profiles for these clusters, see Fig. 2c and e. (e) Boxplots showing the distribution of read-in transcription at 3 to 4 h p.i. for genes in the 50 clusters identified from sense PRO-seq profiles. Boxes represent the range between the first and third quartiles for each cluster. Black horizontal lines in boxes show the median. The ends of the whiskers (vertical lines) extend the box by 1.5 times the interquartile range. Data points outside this range (outliers) are shown as small circles. The red horizontal line indicates the cutoff we previously used to determine that no read-in transcription is observed (≤5% read-in transcription). Metagene plots of PRO-seq…

• 

  FIG 6. HSV-1 infection leads to stronger downstream shifts in pause sites than NELF depletion. Metagene plots around the TSS of PRO-seq profiles for mock and 3-h p.i. WT-F infection from the study of Birkenheuer et al. (5) (a and c) and 0-, 1-, 2-, and 4-h auxin-inducible degradation of NELF from the study by Aoi et al. (47) (b and d) for example clusters showing a broadening of the TSS peak (a and b) or a small downstream peak (c and d) upon NELF degradation. See Materials and Methods and Fig. 1 and 2 legends for the explanation of metagene plots.