RNAPII at Histone Genes in Cancer Diagnosis/Prognosis

March 24, 2025 By Stuart P. Atkinson

MA plots of RELA, PDGFB-1, PDGFB-2, and YAP1 expression. — Mapping hypertranscription spots identifies cCREs and histone genes with elevated tumor signals. MA-style plots rank normalized counts for RELA, PDGFB-1, PDGFB-2, and YAP1. From Henikoff et al.

The Uncertain Link Between Hypertranscription and Cancer

In a recent article in Science (Henikoff and Zheng et al.), researchers from the laboratory of Steven Henikoff (Fred Hutchinson Cancer Center) sought to understand the link between genome-wide hypertranscription a global increase in transcriptional output common in human cancers and poor cancer patient prognosis (Zatzman et al.). The authors hypothesized that hypertranscription might only hold relevance for elements that limit the rapid proliferation of cancer cells and, hence, cancer progression; this prompted the authors to explore the S-phase-specific transcription of the 64 genes that encode the five histone protein subunits required to properly package any newly generated DNA. While previous studies have highlighted the presence of high levels of RNA polymerase II (RNAPII) at these histone genes (Mahat et al.), the unadenylated nature of these histone gene RNAs means that they do not tend to appear in RNA-sequencing-based analyses and, as such, remain somewhat unexplored in this context.

Excitingly, the findings of this new study now suggest that the analysis of RNAPII levels at histone genes may represent a quick and straightforward tool for human cancer diagnosis and prognosis; but can Parallel analysis of individual cells for RNA expression and DNA from targeted tagmentation by sequencing or "Paired-Tag" from Epigenome Technologies take these findings even further? Read on to discover how this single-cell epigenetic technology can contribute to decoding the "silent" mechanisms of diseases such as cancer.

Browser tracks and density peaks for tumor vs normal across tissues. — Tumor-specific chromatin peaks at two 250 kb loci on chr17 show strong ERBB2 and MSL1 signals in breast and colon cancers, and count-density plots across seven tissues highlight ERBB2 and MSL1 hotspots in tumors. From Henikoff et al.

Linking RNAPII Binding to Gene Regulatory Regions and Tumorigenesis

Henikoff and Zheng et al. employed an adapted cleavage under targeted accessible chromatin assay for formalin-fixed paraffin-embedded sections (FFPE-CUTAC; Henikoff et al. 2023) to directly map RNAPII-Ser5p and, as such, measure transcription without the need to analyze RNA, which suffers from problems such as instability and variable transcript half-lives. Initial comparisons at candidate cis-regulatory elements (cCREs) more frequently encountered significantly higher RNAPII levels in mouse tumor tissues than in normal mouse brain tissues. Scaling and normalization supported the sensitive detection of RNAPII hypertranscription, which revealed evident hypertranscription for RELA-driven brain tumors, high and low levels of hypertranscription for PDGFB-driven tumors, and weak hypotranscription in YAP1-driven tumors. Of note, RNAPII abundance differed between tumor samples (high) and normal brain tissues (low) at all classes of regulatory elements (promoters, H3K4me3-marked cCREs, proximal enhancers, distal enhancers, and CTCF cCREs).

Expanding their analysis of cCREs in human samples, the authors discovered clear hypertranscription in five of the seven tumor types assessed (breast, colon, liver, rectum, and stomach); however, the lack of hypertranscription in kidney and lung tumor samples suggested hypertranscription as a common, but not a defining cancer feature. Peak calling via sparse enrichment analysis for CUT&RUN (SEACR; Meers, Tenenbaum, and Henikoff) reported a median of 4483 peaks elevated in tumors and 15 in normal tissue, providing evidence for the common occurrence of RNAPII hypertranscription in human cancer samples. Interestingly, the top 100 cCREs differentially bound by RNAPII in tumor samples overlapped SEACR peaks, with the top-ranked cCREs in breast, colon, liver, lung, and rectum tumors intersecting the MSL1, RFFL, PABPC1, CLTC,SERINC5

Line plots of tumor and normal count densities around loci — Comparison of tumor (red) versus normal (blue) read counts around three genomic intervals (chr14, chr17, chr4) across seven tissue types, revealing tissue-specific enrichment patterns. From Henikoff et al.

RNAPII Levels at Histone Genes Enters the Party

Evaluating whether FFPE-CUTAC could resolve differences between tumor samples employed a cCRE-based UMAP; normal tissue samples produced mixed clusters, while tumor samples formed tight homogeneous clusters separated by tissue type (with few samples and shallow sequencing depths), overall suggesting that RNAPII at regulatory elements better discriminates between tumors than originating tissues. The authors noted that the higher amount of DNA generated thanks to RNAPII hypertranscription and the rapid proliferation of cancer cells might require the elevated expression of S-phase-dependent histone genes; furthermore, the potential for histone production at S-phase to rate-limit proliferation could link cancer aggressiveness and RNAPII levels at histone genes. Indeed, the mouse and human histone S-phase-dependent histone gene analysis initially demonstrated RNAPII hypertranscription in tumor samples. These data agree with previous studies highlighting S-phase-associated high levels of RNAPII at mouse histone genes (Mahat et al.) and at Drosophila histone locus bodies (Lu et al. and Huang et al.).

UMAP construction with FFPE-CUTAC RNAPII cCRE data from meningioma patient samples (non-invasive tumor type) and the previously analyzed tumor/normal pairs revealed that meningiomas clustered separately from other tumors and normal samples; however, the analysis of RNAPII levels at histone genes revealed the clustering of all tumor types, suggesting that this analysis distinguishes cancer from normal tissues but not different cancer types. Attempts to link RNAPII levels to tumor aggressiveness revealed that histone genes performed optimally, which provides additional evidence that high RNAPII levels at histone genes drive proliferation in cancer. Finally, the team assessed if RNAPII levels at histone genes could predict cancer aggressiveness in invasive breast tumors; they discovered that cCREs supported clustering according to tumor type, but histone genes formed a single large cluster that overlapped normal clusters at one distant end, suggesting that histone genes can predict invasive cancer aggressiveness.

UMAP embeddings and WHO grade correlation plot. — UMAPs separate samples by cCRE profiles (A), read depth (B), and 500-bp bin signals (C). Tumors form distinct clusters (D–E), and RNAPII-Ser5p rank correlates with WHO grade. From Henikoff et al.

RNAPII Levels at Histone Genes Combined with Transcriptomics Predicts Tumor Recurrence

The authors integrated FFPE-CUTAC and RNA-sequencing (to evaluate the elevated expression of proliferation genes (Thirimanne et al.) data in the hope of predicting recurrence in meningioma patients and discovered a highly significant association between elevated RNAPII levels at histone genes and rapid patient recurrence. FFPE-CUTAC RNAPII data for histone genes supported the distinct separation of the five most rapidly recurring tumors (from the remaining 25), which agrees with the low recurrence rate of this predominantly benign tumor type. Of note, the authors state that the accurate prediction of poor outcomes by high levels of RNAPII occupancy at histone genes may imply a causal basis.

UMAPs and Kaplan–Meier curves for genomic features. — UMAPs colored by histone gene, ribosomal protein gene, mitochondrial fraction, and chr22q loss highlight tumor–normal separation. High RNAPII or chr22q loss patients show shorter recurrence-free survival. From Henikoff et al.

Histone Overproduction and Cancer-associated Chromosomal Defects

Finally, this exciting study also established a link between high RNAPII levels at histone genes and the loss of whole chromosome arms (a known cancer-associated alteration; Shih et al.) in meningioma and breast tumor samples. The authors hypothesized that the production of large amounts of histone H3 may interfere with the ability of CENP-A - a histone H3 variant to assemble nucleosomes at centromeres and thereby promote the centromere breakage and loss of whole chromosome arms (Giunta et al. and Scelfo et al.). Overall, the data suggest that histone gene hypertranscription induces genomic instability in cancer cells in parallel with driving overproliferation .

Bar chart, cartoon of missegregation, boxplots, and karyotype model. — Frequency of whole-chromosome gains or losses across TCGA cancers, diagrams of centromeric missegregation models, arm-level gain/loss boxplots for acrocentric vs metacentric chromosomes, and a normal-to-tumor karyotype schematic. From Henikoff et al.

Conclusions and Consequences

The authors employ elevated RNAPII levels at genes/regulatory elements to demonstrate cancer-associated genome-wide hypertranscription; furthermore, they observed high RNAPII levels at S-phase-dependent histone genes in tumor cells, which suggests that hypertranscription produces histones for DNA packaging to support rapid cancer cell proliferation and agrees with the hypothesis that histone production rate-limits S-phase progression. Analysis in non-invasive human tumors further underscored this link and demonstrated how RNAPII levels at histone genes could accurately predict rapid recurrence, while a similar analysis revealed how RNAPII levels at histone genes predict aggressiveness in invasive tumor types. The authors suggest that simple assays evaluating RNAPII levels at histone genes of histone gene expression (Sun and Qi) may represent a cheap and straightforward diagnostic/prognostic tool.

This study provides evidence for the utility of measuring RNAPII binding at histone genes in human tumor samples; however, how does RNAPII binding correlate with the chromatin environment surrounding these genes? The profiling of multiple histone modifications combined with simultaneous RNA sequencing at the single-cell level may provide a finer level of potential diagnostic or prognostic data for patients suffering from a range of tumor types. Paired-Tag from Epigenome Technologies generates joint epigenetic and gene expression profiles at the single-cell resolution and detects histone modifications and RNA transcripts in individual nuclei with an efficiency comparable to single-nucleus RNA-seq/ChIP-seq assays. As such, applying Paired-Tag technology may enable cancer researchers to take huge leaps forward in patient diagnosis and prognosis and even aid the choice of therapeutic options. This huge advance was first developed by a team guided by Bing Ren, one of the lead authors of this new study; now, Epigenome Technologies offers optimized Paired-Tag kits and services to researchers in the epigenetics field under an exclusive license from the Ludwig Institute for Cancer Research.

For more on how evaluating RNAPII levels at histone genes may represent a new diagnostic and prognostic tool for human cancer, see Science, February 2025.