The identification and quantification of RNA G-quadruplex (rG4) in vivo is an essential step to study rG4 in cell biology and human diseases. Next-generation sequencing (NGS)-based techniques, such as G4RP-seq [1], BG4 uvRIP-seq, and DMS-seq plus RT Stop profiling[2], have been employed to investigate in rG4 vivo landscapes across the transcriptome and to assess quantitative changes under various conditions. However, all these methods have limitations for accurate rG4 quantifications.

rG4 Array vs G4RP-seq and BG4 uvRIP-seq

In G4RP-seq and uvRIP-seq, samples were cross-linked to preserve transiently formed cellular G4 structures. However, Indirect rG4-RNA interactions may also be cross-linked, thereby potentially leading to false positive results [3].

Furthermore, during their rG4 capture, the binding of BG4 antibody or ligands to rG4 may be obstructed by rG4-binding proteins, such as G3BP1 and nucleolin, which preferentially interact with rG4 structures in vivo and occupy the binding sites. For instance, G3BP1 directly binds to rG4 via its C-terminal RGG domain and modulates mRNA stability[4]. The preoccupation of these proteins to rG4 and their cross-linkage can impede the subsequent binding of BG4 antibody or small-molecule ligands, thereby causing false-negative detection results.

In rG4 Array, the above issue is circumvented by using small chemical DMS methylation masking instead of cross-linking. DMS can readily and swiftly permeate cells and all cellular compartments. It is the in vivo rG4 regions protected from DMS methylation that are refolded for G4 antibody capture and detection for purified RNA devoid of any binding proteins. Arraystar rG4 Microarray profiling detects and quantifies in vivo rG4s through in vivo DMS treatment to mark the native rG4s prior to cell lysis, followed by in vitro rG4 refolding for antibody enrichment, thereby greatly the reliability of quantification of native in vivo rG4 states across the transcriptome.

rG4 Array vs DMS-seq Plus RT Stop Profiling

DMS-seq plus RT stop-profiling approach incorporates in vivo DMS treatment step followed by RT stop profiling. Guanine residues in unfolded rG4 regions is methylated by DMS at their N7 position whereas guanines in folded rG4 structure are not[2]. The guanines protected from DMS methylation by the in vivo rG4 structure can refold into rG4 in vitro, whereas the methylated ones cannot, thus allowing for indirect inference of the rG4 folding state in vivo.

However, DMS can also methylate adenosine (A) at the N1 position and cytosine (C) at the N3 position (Figure 1). In DMS-seq plus RT stop-profiling, the reverse transcriptase stalls and chain-terminates not only when encountering a folded rG4 but also when encountering m1A or m3C. Consequently, the RT could stop at folded rG4 as well as m1A or m3C sites, which can lead to inaccurate quantification of the true rG4 site.

Arraystar rG4 Microarray profiling uses G4-specific antibody BG4 to affinity capture rG4-structure containing RNA fragments, which produces much more authentic rG4 signals than by relying on reverse transcriptase stops. The captured RNAs are then stripped of methylation by AlkB treatment for unimpeded and unbiased reverse transcription for making cRNA and rG4 microarray detection. Thus, the in vivo DMS treatment and in vitro refolding reserve and recover the true native rG4 conformation, together with anti-G4 antibody BG4 to efficiently and specifically capture genuine rG4-containing RNA fragments. Finally, the rG4-RNAs are detected via Arraystar rG4 microarray, much more sensitive than by sequencing. Altogether, Arraystar rG4 array is the most accurate and sensitive platform for in vivo rG4 profiling.

Figure1. DMS-seq plus RT Stop profiling. The reverse transcriptase (RT) stalls and the cDNA chain terminates not only when encountering a folded rG4 but also when encountering DMS methylated m1A or m3C, leading to mis-identification of rG4 sites.

RNA-seq is poor for quantifying rG4 at low-abundance

rG4 formation in living cells (in vivo) is important piece of information in studying rG4. One of the main challenges is the folding and unfolding of rG4s occur rapidly and dynamically in vivo, with the natural equilibrium toward the unfolded state. As a result, the abundance of rG4-containing RNAs in cells is very low.

In RNA-Seq, most sequencing reads come from highly abundant RNAs (e.g. housekeeping genes), while low-abundance RNAs receive low coverage. The low read counts reduce the sensitivity to detect and reliability to quantify the RNA fragments. Increasing coverage depth does not scale linearly to improve the sensitivity and accuracy for low-abundance transcripts: the gain diminishes quickly even with deep RNA-seq coverage. Consequently, the inherently low abundance levels of rG4-RNAs lead to their difficulty in detection and poor quantification by RNA-seq based methods.

For example, lncRNA-MALAT1, a well-studied RNA for its rG4 formation, has faced this limitation in Global mapping of RNA G-quadruplexes by sequencing (G4RP-Seq) [3]. That is, G4RP-Seq requires adding G4-stabilizing ligands BRACO-19 and RHPS4 that can artificially induce and stabilize G4s to detect significant rG4 signals from MALAT1[1]. However, it is not ideal for observing natural, transient rG4s under normal physiological conditions.

Figure2. MALAT1 rG4 signals from G4RP-Seq for biotin control, untreated, BRACO-19-treated, and RHPS4-treated samples [3]. While treatment with G4-stabilizing ligands BRACO-19 or RHPS4 produces high rG4 signals, the rG4s are artificially induces and stabilized G4s [1], not representing the native transient rG4 states in vivo and not ideal for observing the natural rG4s under normal physiological conditions.

In addition, in RNA sequencing (RNA-seq), RPKM (Reads Per Kilobase of transcript per Million reads mapped) is used to normalize for differences in both library size and gene length. However, the RPKM value reflects the relative abundance of a transcript within that sample, not its quantity. Comparing transcript expression levels across different samples using RPKM can be misleading because variations in RNA composition affect the overall RNA profile. These differences can distort the RNA expression levels when making cross-sample comparisons. Therefore, rG4 profiling by RNA-seq is not suitable for comparing rG4 abundance between different samples.

Using rG4 microarray profiling, an RNA target is hybridized with its sequence-specific probe, independent of other RNA sequences present even at high abundance. Whereas high abundance RNAs (e.g. housekeeping gene RNAs) can displace low abundance RNA coverage in RNA sequencing, they have little to no impact on low-abundance transcript detection on microarray. Consequently, microarray-based profiling is particularly well-suited for quantifying rG4-RNAs that are characteristically expressed at low levels.

Related Service

Arraystar rG4 Microarray Service

 

Reference

1. Yang SY et al: Transcriptome-wide identification of transient RNA G-quadruplexes in human cells. Nat Commun 2018, 9(1):4730.[PMID: 30413703]
2. Guo JU, Bartel DP: RNA G-quadruplexes are globally unfolded in eukaryotic cells and depleted in bacteria. Science 2016, 353(6306).[PMID: 27708011]
3. Yang SY, Monchaud D, Wong JMY: Global mapping of RNA G-quadruplexes (G4-RNAs) using G4RP-seq. Nat Protoc 2022, 17(3):870-889.[PMID: 35140410]
4. He X, Yuan J, Wang Y: G3BP1 binds to guanine quadruplexes in mRNAs to modulate their stabilities. Nucleic Acids Res 2021, 49(19):11323-11336.[PMID: 34614161]
5. Jiang L et al: Synthetic spike-in standards for RNA-seq experiments. Genome Res 2011, 21(9):1543-1551.[PMID: 21816910]
6. Labaj PP et al: Characterization and improvement of RNA-Seq precision in quantitative transcript expression profiling. Bioinformatics 2011, 27(13):i383-391.[PMID: 21685096]
7. Toung JM, Morley M, Li M, Cheung VG: RNA-sequence analysis of human B-cells. Genome Res 2011, 21(6):991-998.[PMID: 21536721]
8. Kretz M et al: Suppression of progenitor differentiation requires the long noncoding RNA ANCR. Genes Dev 2012, 26(4):338-343.[PMID: 22302877]
9. Xu W et al: Human transcriptome array for high-throughput clinical studies. Proc Natl Acad Sci U S A 2011, 108(9):3707-3712.[PMID: 21317363]