|Home | About | Journals | Submit | Contact Us | Français|
Long noncoding RNAs are key regulators of chromatin states for important biological processes such as dosage compensation, imprinting, and developmental gene expression 1,2,3,4,5,6,7. The recent discovery of thousands of lncRNAs in association with specific chromatin modification complexes, such as Polycomb Repressive Complex 2 (PRC2) that mediates histone H3 lysine 27 trimethylation (H3K27me3), suggests broad roles for numerous lncRNAs in managing chromatin states in a gene-specific fashion 8,9. While some lncRNAs are thought to work in cis on neighboring genes, other lncRNAs work in trans to regulate distantly located genes. For instance, Drosophila lncRNAs roX1 and roX2 bind numerous regions on the X chromosome of male cells, and are critical for dosage compensation 10,11. However, the exact locations of their binding sites are not known at high resolution. Similarly, human lncRNA HOTAIR can affect PRC2 occupancy on hundreds of genes genome-wide 3,12,13, but how specificity is achieved is unclear. LncRNAs can also serve as modular scaffolds to recruit the assembly of multiple protein complexes. The classic trans-acting RNA scaffold is the TERC RNA that serves as the template and scaffold for the telomerase complex 14; HOTAIR can also serve as a scaffold for PRC2 and a H3K4 demethylase complex 13.
Prior studies mapping RNA occupancy at chromatin have revealed substantial insights 15,16, but only at a single gene locus at a time. The occupancy sites of most lncRNAs are not known, and the roles of lncRNAs in chromatin regulation have been mostly inferred from the indirect effects of lncRNA perturbation. Just as chromatin immunoprecipitation followed by microarray or deep sequencing (ChIP-chip or ChIP-seq, respectively) has greatly improved our understanding of protein-DNA interactions on a genomic scale, here we illustrate a recently published strategy to map long RNA occupancy genome-wide at high resolution 17. This method, Chromatin Isolation by RNA Purification (ChIRP) (Figure 1), is based on affinity capture of target lncRNA:chromatin complex by tiling antisense-oligos, which then generates a map of genomic binding sites at a resolution of several hundred bases with high sensitivity and low background. ChIRP is applicable to many lncRNAs because the design of affinity-probes is straightforward given the RNA sequence and requires no knowledge of the RNA's structure or functional domains.
Design anti-sense DNA tiling probes for selective retrieval of RNA target by ChIRP.
Collect cells that will be used for ChIRP experiment.
Crosslink collected cells with glutaraldehyde to preserve RNA-Chromatin interactions and prepare cell pellet.
Lyse crosslinked cells to prepare cell lysate.
Shear DNA by sonicating crosslinked cell lysates.
Hybridize biotinylated DNA probes to RNA and isolate bound chromatin.
Extract RNA fraction from ChIRP samples to quantitate by qRT-PCR.
Extract DNA fraction from ChIRP samples to identify by sequencing or quantitate by qPCR.
Figure 1 depicts the ChIRP workflow. A successful ChIRP experiment typically enriches target RNA significantly over non-specific RNAs. Figure 2 shows enrichment of human telomerase RNA (TERC) from HeLa cells over GAPDH, an abundant cellular RNA that serves as a negative control. Majority of TERC RNAs (~88%) present in the cell were pulled down by performing ChIRP, whereas only 0.46% of GAPDH RNA was retrieved, demonstrating an enrichment factor of ~200 fold. Nonspecific probes such as probes targeting LacZ RNA, which is not expressed in mammalian cells (Figure 2), can be used as additional negative controls.
DNA regions expected to bind the target lncRNA are typically enriched over negative regions when measured by qPCR. Figure 3 shows qPCR validation of four HOTAIR-bound sites in primary human foreskin fibroblasts that we determined by performing ChIRP-seq in the same cell line, while TERC and GAPDH DNA sites serve as negative control regions. Both "even" and "odd" probe sets yielded comparable enrichment of expected HOTAIR-bound sites over negative regions, a hallmark of true lncRNA-binding sites.
High-throughput sequencing of ChIRP enriched DNA yields a global map of lncRNA-binding sites. The Drosophila lncRNA roX2 is known to interact with the X-chromosome in a manner that is required for dosage compensation. Figure 4 shows roX2 binding profile over a section of the X chromosome. Both "even" and "odd" samples have been sequenced and their unique noises have been eliminated to produce a track of overlapping signals. Each "peak" here indicates a strong site of roX2 binding. The complete track and list of roX2 target genes have been described in Chu et al. 2011 17.
Figure 1. Flow chart of the ChIRP procedure. Chromatin is crosslinked to lncRNA:protein adducts in vivo. Biotinylated tiling probes are hybridized to target lncRNA, and chromatin complexes are purified using magnetic streptavidin beads, followed by stringent washes. We elute lncRNA bound DNA or proteins with a cocktail of Rnase A and H. A putative lncRNA binding sequence is schematized in orange. Previously published in Chu et al. 2011.17
Figure 2. ChIRP enriches for human TERC RNA. TERC-asDNA probes retrieve ~88% of cellular TERC RNA and undetectable GAPDH. LacZ-asDNA probes are used as negative controls and retrieve neither RNAs. Mean + s.d. are shown. Previously published in Chu et al. 2011.17
Figure 3. HOTAIR ChIRP-qPCR in primary human foreskin fibroblasts. NFKBIA, HOXD3-4, SERINC5 and ABCA2 are regions that interact with HOTAIR. TERC and GAPDH served as negative controls. Mean + s.d. are shown. Previously published in Chu et al. 2011.17
Figure 4. ChIRP-seq data of roX2 RNA in Sl2 Drosophila cells. "Even" and "odd" were sequenced separately; their data merge to reflect only common peaks in both. The merged track is shown. Previously published in Chu et al. 2011.17
Here we described ChIRP-seq, a method of mapping in vivo lncRNA binding sites genome-wide. The key parameters for success are the split pools of tiling oligonucleotide probes and glutaraldehyde crosslinking. The design of affinity-probes is straightforward given the RNA sequence and requires no prior knowledge of the RNA's structure or functional domains. Our success with roX2, TERC, and HOTAIR - three rather different RNAs in two species - suggests that ChIRP-seq is likely generalizable to many lncRNAs. As with all experiments, care and proper controls are required to interpret the results. Different lncRNA may require titration of conditions, and judicious change of conditions, such as selection of different affinity probes or crosslinkers, may highlight different aspects of RNA-chromatin interactions. Like ChIP-seq, not all binding events are necessarily functional, and additional studies are required to ascertain the biological consequences of RNA occupancy on chromatin. Nonetheless, we foresee many interesting applications of this technology for researchers of other chromatin-associated lncRNAs, which now number in the thousands8,9. Just as ChIP-seq has opened the door for genome-wide explorations of DNA-protein interactions, ChIRP-seq studies of the "RNA interactome" may reveal many new avenues of biology.
C. Chu and H.Y. Chang are named as inventors on a patent application based on this method.
We thank T. Hung, MC. Tsai, O. Manor, E. Segal, M. Kuroda, T. Swigut, and I. Shestopalov for discussions. Supported by the Agency of Science, Technology and Research of Singapore (C.C.), NIH R01-CA118750 and R01- HG004361 (H.Y.C.), and California Institute for Regenerative Medicine (H.Y.C.). H.Y.C. is an Early Career Scientist of the Howard Hughes Medical Institute.