Reversing the Central Dogma: RNA guided control of DNA in epigenetics and genome editing

SUMMARY

The central dogma of the flow of genetic information is arguably the crowning achievement of 20^th century molecular biology. Reversing the flow of information from RNA to DNA or chromatin has come to the fore in recent years, from the convergence of fundamental discoveries and synthetic biology. Inspired by the example of long noncoding RNAs (lncRNAs) in mammalian genomes that direct chromatin modifications and gene expression, synthetic biologists have repurposed prokaryotic RNA-guided genome defense systems such as CRISPR to edit eukaryotic genomes and epigenome. Here we explore the parallels of these two fields and highlight opportunities for synergy and future breakthroughs.

eTOC blurb

Here, Chang and Qi discuss reversing the flow of information in the Central Dogma, from RNA to DNA or chromatin, highlighting lncRNAs and prokaryotic RNA-guided genome defense systems such as CRISPR to edit eukaryotic genomes and epigenome. They explore the parallels between the two and the future of RNA-based technologies.

INTRODUCTION

The central dogma of gene expression is stunning in its explanatory power and ubiquity in biological systems. As a fundamental theory in molecular biology, the central dogma states that genetic information flows in one direction, from DNA, to RNA, to protein. From a historical context, the discovery of reverse transcriptase was one the first major revision to the idea of unidirectional information flow from DNA to RNA to protein. In the first part of the 21st century, discovery of small regulatory RNAs was initially focused on post-transcriptional regulation. Subsequent studies of PIWI-interacting RNAs (piRNAs) and long noncoding RNAs (lncRNAs) led to the recognition of the potential of RNA to influence DNA—specifically changing DNA methylation or chromatin in a heritable fashion¹. In bacteria, retrons are genetic components composed of reverse transcriptase (RT) and non-coding RNA (ncRNA). The RT uses the ncRNA as a template to generate a chimeric RNA/DNA molecule with covalently bonded RNA and DNA components, which work as defenses against phages². The discovery of CRISPR greatly simplified the task of targeting specific DNA sequences³ and led to a flowering of ideas and new strategies to expand the programmability of RNA, with the aid of protein factors, to control the information flow backwards to DNA (Figure 1). While proteins are considered major players on controlling gene expression, these mounting examples highlight RNA as the often-overlooked and active ingredient in reversing the central dogma.

Figure 1.

Diagram showing scheme for reversing the Central Dogma. (A) Genetic information flows from DNA to RNA to protein. Natural and synthetic systems allow reversing this flow direction by modifying DNA based on information encoded in RNA, with examples including piRNA, lncRNA, and CRISPR. (B) Examples for lncRNA and CRISPR for targeting genome regions and interacting with host factors or synthetic regulators to modulate gene expression, chromatin status, etc.

LONG NONCODING RNAS: INFORMATION FLOW FROM RNA TO CHROMATIN

Eukaryotic genomes are pervasively transcribed to produce numerous lncRNAs. LncRNAs are defined as transcripts that are greater than 200 nucleotides, may be spliced or unspliced, and apparently do not function by encoding for a protein or peptide. This definition of exclusion arose during an era of a protein-centric view of the genome and genome assemblies, despite the fact that just 2% of the human genome is coding. The discovery, characterization, function of lncRNAs have been extensively reviewed elsewhere⁴ and we will focus on salient examples that facilitate our discussion. In brief, the human genome encodes over 60,000 lncRNAs, three times the number of protein coding genes. Inherited mutations in lncRNA genes cause Mendelian disorders ⁵ such as developmental malformations ⁶.Unbiased genome-wide association studies revealed that inherited variation of lncRNA loci in the human population underlies over one thousand trait associations, 800 of which were not explained by neighboring protein-coding genes⁷. For instance, lncRNA quantitative trait loci are linked to inflammatory bowel disease, type 1 and type 2 diabetes, and coronary artery disease, as well as rare variant associations to body mass index. Systematic functional screens in cells show that approximately 2% of lncRNA gene loci are required for cell growth or survival, the vast majority in a tissue or cell-type selective function⁸, which is concordant with more limited examples in genetically engineered animal models.

Many lncRNAs have their functions in the nucleus, recognized archetypes of signal guides, scaffolds, and decoys⁹. First, lncRNAs are produced by the act of transcription, and thus lncRNA marks its chromosomal allele in cis as transcriptionally active. Classic examples of imprinting, where paternal vs. maternally inherited alleles are differentially expressed, are often associated and regulated in part by allele-specific transcription of lncRNA ^10,11. Second, lncRNAs can serve as guides to target chromatin modification machinery. Most if not all chromatin modifying enzymes lack intrinsic DNA binding specificity and rely on RNA or transcription factors to guide them to the target DNA loci. The mechanism of lncRNA targeting is diverse and may be limited to the local genome neighborhood or chromosome from which it is transcribed (in cis) or broadened to loci throughout the genome (in trans) ⁵. Reported target mechanisms include RNA: DNA: DNA triple formation^12,13, lncRNA RNA duplex interaction with nascent RNA ¹⁴, and exploitation of 3D spatial proximity^15,16. Third, lncRNAs are comprised of multiple modular domains that create a scaffold for the assembly of multiple chromatin modification activities. The lncRNA Xist beautifully illustrates each of these archetypes¹⁷. In each female cell in mammals, one of two X chromosomes is transcriptionally silenced by action of Xist. Xist is only transcribed from the inactive X chromosome (Xi), and Xist RNA spreads across the Xi to nucleate and spread dozens of RNA binding proteins and chromatin regulators to silence gene expression through the lifetime of that cell and all its progenies. This classical example of lncRNA and many other recent entries highlight Nature’s heretical invention and elaboration of using RNA to alter and dictate information flow from DNA.

RNA GUIDED DNA TARGETING

The discovery that prokaryotes possess a form of RNA-targeted immunity, termed clustered regularly interspaced short palindromic repeat (CRISPR), has dramatically expanded the view on the physiological role played by noncoding RNAs ^18–20. Complexed with CRISPR-associated (Cas) proteins, these noncoding RNAs, termed CRISPR RNA (crRNA), serves as guides to direct Cas proteins to specific DNA sequences for DNA cleavage ^21,22. This provides an unprecedented example for reversing the Central Dogma, -- how information encoded in RNA can alter information in DNA. In this review, we focus on how RNAs are leveraged for reversing the central dogma in eukaryotes.

Cas9 and Cas12

There are two distinct classes of CRISPR systems²³, and Class II enzymes have their targeting and nuclease functions encoded in a single protein (including Cas9) which span across a large group of RNA-guided nucleases that have evolved in various prokaryotic species. The most used single Cas effectors are Cas9 and Cas12a (formerly known as Cpf1) ^24,25.

Evolution in different host environments have endowed Cas enzymes with unique traits, but they all share a common feature: formation of a ribonucleoprotein (RNP) for DNA or RNA targeting. For example, all Cas9 enzymes require complexation with two non-coding RNAs: one CRISPR-associated RNA (crRNA) encoding the target DNA complementarity and one trans-acting crRNA (tracrRNA) ²⁶. On the contrary, all Cas12a enzymes utilize a single crRNA without tracrRNA. This can be explained by the fact that Cas12a encodes RNase activity that processes the precursor crRNA array to individual crRNAs for RNP formation, while Cas9 requires tracrRNA and host factors (e.g., RNase III) for the same process ²⁷. Nevertheless, the evolution of tracrRNA is an interesting event, which may endow unique features, such as offering a versatile evolutionary solution to process long pre-crRNA arrays to individual crRNAs, while maintaining the compactness of the crRNA array. To simplify the use of two RNAs, crRNA and tracrRNA are fused as a single guide RNA (sgRNA).

All DNA-targeting Cas nucleases require a protospacer adjacent motif (PAM), short DNA segments flanking the targeted genomic locus that are not encoded in the corresponding region flanking the crRNA ²⁸. PAM sites provide an entry point for the Cas-guide RNA complex for DNA binding and melting. PAMs from different Cas9 or Cas12 species have varied stringency. Examples include that, Streptococcus pyogenes Cas9 requires a NGG PAM, Staphylococcus aureus Cas9 recognizes an NNGRRT PAM (N is any nucleotide and R is A or G), and some Cas12a molecules use a TTTV PAM (V is A or C or G) ^25,29,30. Protein engineering of the PAMbinding domain of the Cas proteins can alter the PAM preference, which is an important factor for determining its targetable genomic space ^{31 32 33 34}. Structural and mechanistic insights of the sgRNA have also improved CRISPR-mediated gene editing. For example, varying the full-length of tracrRNA may improve the editing activity^{35 36}. Truncated sgRNAs that guide Cas9 binding but not cleavage can shield off target sites and prevent off-target cleavage without affecting on-target sites^37. Similarly, introduction of RNA hairpins to stabilize sgRNA can increase the specificity of cleavage of Cas9 and Cas12a^38.

Engineered guide RNA with aptamers for instructions

SgRNA can be universally engineered into a scaffold that recruits diverse enzymatic functions. In this case, an sgRNA sequence can be fused to a protein-interacting RNA aptamer that recruits specific RNA-binding proteins (RBPs), termed scaffold RNAs (scRNAs)^39,40. RBPs fused with activators or repressors can then be used to bring additional instructions onto the scRNAs. When orthogonal aptamer–RBP pairs (e.g., MS2–MCP, PP7–PCP) are used to generate different scRNAs, distinct regulatory modules can be recruited to different DNA sites for distinct regulation. These scRNAs mimic the scaffolding natural function of lncRNAs that can encode both the target location and independent mode of regulatory action. Several systems for enhanced CRISPR-mediated gene regulation leverages this approach, including the broadly used SAM system that fuses MS2 aptamers to the sgRNA structure and recruits multiple activator domains to enhance target gene regulation⁴¹

Engineering guide RNA with additional RNA domains is a powerful approach to expand CRISPR functionality on the DNA. In addition to DNA regulation, the method of incorporating MS2 or PP7 RNA aptamers into the sgRNA has been utilized for multiple-color chromatin labeling, wherein RNA aptamers recruit MS2 or PP7 coat proteins fused with different fluorescent proteins to various target genomic loci ⁴². In a method termed CRISP-Display, large RNA motifs including natural lncRNAs was directly inserted into the CRISPR sgRNA, allowing for targeting functional RNAs and associated ribonucleoprotein (RNP) complexes to desired genomic loci⁴³. Young and colleagues tethered a short RNA (60-nt RNA derived from the promoter sequence of Arid1a) to the sgRNA using CRISPR-Cas9 in the vicinity of YY1 binding sites and determined the tethered RNA specifically increased the occupancy of YY1 at nearby enhancers⁴⁴. Thus, engineering guide RNA by coupling with other RNA domains offers a scaffolding system that enables multimodal functionality on target genes using a single Cas protein.

Compact Cas molecules

Recently identified CRISPRs revealed new compact Cas effectors. Specifically, Cas12f (also called Cas14, 400–700 amino acids) and Cas12j (also called CasΦ, 700–800 amino acids) have been characterized from Archaea and bacteriophages^45,46 . In the case of Cas12f, two Cas12f molecules form a homodimer which binds to one crRNA-tracrRNA complex as a RNP ⁴⁷. This reduces the ‘genetic cost’ of expression and delivery, as expressing Cas12f and its associated crRNA-tracrRNA takes a smaller fragment size. While the wildtype system showed limited performance in mammalian cells, protein engineering was used to create better performing Cas variants in mammalian systems. One example is our development of a modified Cas12f system (termed CasMINI), which is 1.6 kb and less than half the size of Cas9 ⁴⁸. Notably, sgRNA engineering by altering the scaffold sequence played an important role in enhancing the activity of engineered Cas12f systems, which is likely due to better RNP formation and stability.

The compact size of miniature Cas proteins offers useful solutions for better delivery, using adeno-associated virus (AAV) or lipid nanoparticles (LNP), and promise better in vivo applications. Interestingly, large Cas effectors such as Cas9 or Cas12a (3~4kb) often utilizes smaller crRNA (or crRNA-tracrRNA)²⁹, and smaller Cas effectors such as Cas12f (1.2~1.8kb) uses larger crRNA complexes⁴⁹. This makes sense structurally as smaller proteins require larger noncoding RNAs, serving as ‘ligand’, to form a sufficiently stable ribonucleoprotein for interaction with the stable double-stranded DNA.

Base editors

RNA-guided Cas nucleases induce double-strand break (DSBs) in DNA and DSB-mediated random indels (insertion/deletion), which is a ‘primary’ form of RNA->DNA information writing using CRISPR. Beyond this, Cas molecules can be fused to nucleotide modifiers to enable other forms of DNA editing. DNA base editing enzymes have been fused to the nuclease-dead dCas or the nickase variant of Cas that generates a single-strand break, to generate precise DNA mutations at a targeted location. Noncoding RNA transcripts guide activation-induced cytidine deaminase to mediate somatic hypermutation in B cells ⁵⁰, providing a precedent for using RNA to guide base editing.

The cytidine deaminase enzyme APOBEC1, which converts cytosine (C) to uracil (U), was fused to nCas9. This base editor, termed cytosine base editors (CBEs), catalyzes a targeted C-to-U on the DNA strand not bound by sgRNA, which is read as a thymine (T) upon DNA replication and creates a precise C to T mutation ^51–53. Fusing uracil DNA glycosylase (UGI) and mutating APOBEC1, enhances the C-to-T reaction. In a similar manner, nCas9 fused with an engineered Escherichia coli TadA, termed adenosine base editors (ABEs), created an adenosine (A) to guanine (G) conversion ^54–56. ABE was also engineered into an efficient cytosine base editor ⁵⁷. Fusing both cytosine and adenosine deaminases to nCas9 generates both C-to-T and A-to-G conversions at a single site ^58,59. Base editors for nucleotide transversion, conversion between a purine and a pyrimidine, have been developed for C-to-G and C-to-A ^60–62. In addition to nCas9, the nuclease-dead Cas12a can efficiently modulate base editing ⁶³. Base editors are great examples for how RNAs are used as instructions to introduce specific mutations at DNA regions.

RNA AS GENOME TEMPLATE

Beyond RNA as a guide, RNA can directly template DNA sequence addition and connectivity. In yeast Saccharomyces cerevisiae, endogenous transcript RNA is used to mediate homologous recombination with chromosomal DNA ^{64 65}. In eukaryotic cells, telomerase RNA is the scaffold for the assembly of the multiprotein telomerase enzyme complex that protects the ends of linear chromosomes (reviewed by ⁶⁶). Moreover, telomerase RNA provides the template for the addition of hundreds to thousands of a precise repeat sequence to chromosome ends by the reverse transcriptase activity of TERT. Ciliates, like the unicellular organism Oxytrichia, undergo a highly unusual life cycle that involves fragmenting its germline genome into numerous small pieces, eliminating ~95% of its genome, and then assembling the remaining fragments into ~16,000 genesized chromosomes. The assembly into chromosomes is not random but has been shown to be guided by RNA ^67,68—that is the order of genes on the RNA specifies the order of the DNA pieces that are stitched together ⁶⁹. Current model suggests base pairing between the RNA and regions of DNA sequences. However, because point mutations in the RNA are not copied into the assembled chromosomal DNA, the Oxytrichia lncRNA is likely not a template that is reverse transcribed into the DNA. Nonetheless, these examples illustrate the possibility of RNA to program long sequences into DNA by either direct or indirect means.

Prime editors

To further expand the mode of DNA writing, Liu and colleagues developed a method termed prime editing to insert longer stretches of DNA ^70–72 Click or tap here to enter text.. In this method, nCas9 is fused to a reverse transcriptase and the sgRNA is engineered with an elongated 3’ sequence that encodes the desired mutations and a priming region complementary to the DNA. This altered sgRNA is a critical development, serving as a template encoding multiple instructions (both target location and desired mutations), which is collectively called prime editing guide RNA (pegRNA). In this RNA-directed system, the pegRNA primes the nicked DNA strand and the reverse transcriptase converts the template sequence from RNA to DNA, creating a targeted insertion or deletion in place. It was utilized to make all 12 base-to-base conversions, as well as small-fragment insertions and deletions (<100 bp). Combining with Bxb1 recombinase, the pegRNA system enables larger insertion or deletions (up to several kb) ^74,75.

Notably, several improvements in prime editing have taken place via iterative improvement of the pegRNA design. Nelson et al. found that degradation of the 3’ template end of pegRNA appears to be a major bottleneck ⁷³. Such truncated pegRNAs cannot serve as template for RT, and moreover may block productive full length pegRNA-CasRT complexes from occupying target sites in a dominant negative manner. Enhancing greater stability of linear pegRNAs has improved the prime editing. Liu et al. further found that pegRNA can be split into a guide and a template RNA, and greatly stabilized by circularization, which makes the template RNA impervious to exonucleases ⁷⁶. Circular template RNA appears to enable greater duration of productive enzyme occupancy of the target site DNA, because dCas9 and RT can be expressed as separate proteins and still support prime editing. This split design solves the size problem of the Cas9-RT fusion construct, a current limitation for AAV delivery of genetic medicines in vivo ⁷⁷.

RNA-guided DNA transposition

More recently, diverse Cas proteins were discovered to exhibit targeted genomic transposition or insertions. For example, A Type I-F and a Type II-K CRISPR system can insert DNA fragments to a genomic site via RNA-guided CRISPR-mediated recruitment of transposition machinery ^78,79. This ability for targeted transposition has been demonstrated in vitro and in prokaryotic hosts ⁷⁸. While targeted transposon needs to be developed in human cells, these offer excellent examples on how CRISPR can use RNA instructions to enable diverse forms of DNAlevel information rewriting. Beyond natural Cas-transposase systems, Durrant et al. repurposed and fused bacterial serine integrase with dCas9 to enable programmable insertion with multikilobase DNA inserts ⁸⁰ .

RNA-GUIDE GENE REGULATION AND EPIGENETIC EDITING

CRISPRa

What best illustrates the concept of reversed RNA->DNA information flow is probably the development of CRISPR activation (CRISPRa) ^81,82. In this method, a nuclease-dead dCas (e.g., dCas9, dCas12a, dCasMINI) is fused to a transcriptional activator ^83–85. Guided by its cognate sgRNA bearing a sequence indicating the target DNA locus, the RNP can specifically bind to an endogenous genomic locus to switch on the gene that is nearby. In mammalian cells, various transactivator domains such as VP64, tripartite VPR (VP64, p65, and RTA) are fused to dCas9 ⁸⁶, tandem repeats of peptides ⁸⁷, or indirect recruitment via engineered RNA scaffold, have been adopted for gene activation ⁸⁸. In bacterial cells, fusing synthetic transcriptional activators or the ω subunit of RNA polymerase to dCas9 enabled gene activation ^89,90. Unlike ectopic transgene expression, CRISPRa can be used to precisely tune the magnitude of gene upregulation. Furthermore, CRISPRa can be utilized to upregulate multiple gene targets by simply using multiple sgRNAs. In some cases, via a dCas12a-activator system, a compact crRNA array that can encode multiple gene-site instructions to activate many genes (7~10) can be achieved ^91,92.

CRISPRi and CRISPRoff

CRISPR interference (CRISPRi) allows programmable gene silencing and is the conceptual opposite of CRISPRa. In CRISPRi, dCas9 is fused to a protein domain, such as KRAB, that recruits the endogenous SETDB1 histone modification enzyme and additional factors that places the repressive chromatin mark histone H3 lysine 9 trimethylation (H3K9me3) ⁹³. CRISPRi deposits H3K9me3 on ~1 kb window surrounding the target locus and can be targeted to promoters and distal enhancers in a multiplexed fashion to study gene or DNA element function ^94,95. One fruitful application of CRISPRi was the systematic functional interrogation of human lncRNA genes ⁹⁶, and the factors required for maintenance of XIST-mediated gene silencing in adult somatic cells ⁹⁷.

CRISPRoff is a new generation of programmable gene silencing with epigenetic memory. CRISPRoff builds on CRISPRi and contains dCas9, KRAB domain, and is further fused to DNMT3A and DNMT3L, components of the de novo DNA methyltransferase complex ⁹⁸. This combination of DNA element deactivation (KRAB) and maintenance (DNMT3A/3L) mimics the natural mechanism of epigenetic genetic silencing, such as X inactivation. CRISPRoff silences gene expression from the targeted promoters or enhancers, and further adds DNA methylation to the target locus. DNA methylation is copied and propagated by endogenous DNMT1 during cell division, thus ensuring memory of gene silencing over multiple somatic cell generations or even cell differentiation ⁹⁹. CRISPRoff allows a “hit-and-run” approach to epigenetic editing. Unlike CRISPRa or CRISPRi that needs to be continually expressed to exert their effects, transient expression of CRISPRoff construct and sgRNA allows durable gene silencing in cultured cells. The application of CRISPRoff in vivo and primary cells is an active area of investigation.

RNA guided epigenetic editing

Fusion of dCas with epigenetic domains with different enzymatic activities can be used to write or erase a broad spectrum of chromatin chemical modifications. To erase DNA methylation, TET1 was fused with dCas9 to remove DNA methyl groups and upregulate gene expression at specific chromatin sites ^99,100. In addition, to specifically add acetyl groups to local histones, the catalytic core of the acetyltransferase p300 domain was fused to dCas9 ^101–103. When directed to enhancers, this fusion adds an acetyl group to Lysine 27 of Histone H3 (H3K27ac), leading to activation of gene expression. Other modes of writing methylation at H3K4, H3K9, and H3K27 have been achieved by engineering the fusion effectors to dCas proteins ¹⁰⁴. All of these expand the modes that RNA->DNA epigenetic changes.

IN THE HORIZON

Here we have discussed several parallel themes in lncRNA biology and CRISPR technologies. What are potential areas of cross pollination and advances in the future? We offer several suggestions that are by no means exhaustive. Indeed, we look forward to being surprised by the richness of biology and the ingenuity of investigators in the field.

Gene activation with memory

While CRISPRoff has achieved gene silencing with epigenetic memory, no comparable technology exists for durable gene activation. CRISPRa requires continual expression of the dCas9-VP64 construct. The ability to durably induce gene expression after a transient expression of a synthetic regulator will offer many advantages for experimental biology and medicine. For instance, genetic haploinsufficiency can be treated by durable gene activation. LncRNA biology offers case studies for how durable gene activation is achieved in nature. Dosage compensation in Drosophila involves the transcriptional activation of the single male X chromosome by two-fold via two lncRNAs named roX1 and roX2 ^38,105.The roX RNAs are also transcribed from the X chromosome and guide a protein complex that hyperacetylates histones and promotes transcriptional elongation on the X, creating a positive feedback loop. In the mammalian HOXA locus, the lncRNA HOTTIP is located on the 5’ end of the locus, associates with the MLL H3K4 methylase complex and promotes expression of the 5’ HOXA genes, including itself ¹⁰⁶. Both dosage compensation and positional identity are both remarkably stable once established and can even persist despite extensive ex vivo culture ^107,108. Accessing endogenous systems of positive feedback loops with cis-acting lncRNAs may unlock programmable gene activation with greater durability.

RNA control 3D genome and nuclear architecture

LncRNAs exert powerful roles in the compartmentalization of nuclear neighborhoods (reviewed by ^109,110) (Figure 2). The nucleolus (rRNA repeats), nuclear paraspeckle (Neat1), and the Barr body (Xist) are several prominent examples. In X inactivation, Xist not only spreads across the inactive X chromosome (Xi), Xist also compacts and pulls in the Xi and localizes the Xi to the nuclear periphery ¹¹¹, creating a chromosomal structure and nuclear territory that are distinct from that of the homologous active X chromosome. Synthetic biologists have started to use RNA to program the 3D genome. CLOUD9 is a system that targets multiple dCas9 proteins fused to inducible dimerization domains to DNA elements spaced kilobases apart ¹¹². Upon sgRNA expression and dimerization, distal DNA elements such as enhancers and promoters can be pulled into spatial proximity, leading to gene activation ¹¹². CRISPR-GO is a method to direct chromosomal loci position within the nucleus. Via a dimerization system that repositions the target DNA to desired nuclear compartment, CRISPR-GO mediates RNA-guided spatial relocalization of specific chromatin to a different location including nuclear periphery, Cajal bodies, or heterochromatin condensates ^113–115. The apparent difference is the scale and complexity of the 3D genome architecture being programmed by endogenous lncRNAs. One potential mechanistic distinction is that endogenous lncRNAs have much more elaborate scaffold function, and each unit of the scaffold is repeated multiple times. Xist is an 81 kb lncRNA in the mouse and consists of multiple repeats, named A-F, which have a division of labor in spreading across the chromosome, gene deactivation, and epigenetic memory ^116,117.

Figure 2.

Future direction for using RNA to modulate DNA 3D configuration and phase condensation for using lncRNA (A) or CRISPR (B).

RNA induced condensates

Nuclear RNAs are powerful nucleators and participants in multivalent RNA-protein interactions. Studies of condensates to create membraneless compartments have relied extensively on in vitro biochemistry of purified component proteins and imaging in cells. Tools to program condensate formation and dissolution are starting to be developed¹¹⁸. LncRNAs offer some potentially useful lessons. The lncRNA NORAD contains 18 copies of binding sites for the RNA binding protein Pumilio and titrates Pumilio activity in the cell. NORAD can compete Pumilio away from the multitude of Pum binding sites in the mRNA transcriptome, even though the latter collectively exists in ~40-fold excess ¹¹⁹. Mendell and colleagues showed that the repeated nature of Pum binding sites in NORAD creates multivalency that facilitates Pumilio condensate formation in a super-stochiometric manner ¹¹⁹. A synthetic circular RNA composed of Pum binding sites phenocopied NORAD function. These emerging studies suggest the possibility that elaborating and engineering scaffolds on sgRNAs targeting DNA or RNA may be a fruitful path to new technologies to dissect and manipulate condensates and complex multivalent molecular interactions.

The 21th century has been envisioned as the century of biology. Rewiring the the central dogma with RNA may play major roles in the next phase of exciting biological advances.