Circular RNA

For a long time, RNA therapeutics followed a fairly straight path. First came antisense oligonucleotides, then siRNA, and more recently in vitro transcribed linear mRNA. The success of the mRNA COVID-19 vaccines did more than validate a single product. It stress tested the entire RNA ecosystem. Manufacturing scaled at record speed. Lipid nanoparticle delivery matured. Regulators gained real-world experience with an RNA platform. It felt like a coming of age moment for the field.

Still, linear mRNA has its limitations. One of the core tensions is durability versus immunogenicity. You want strong, sustained protein expression, but cells are built to detect foreign RNA. Structural features, double stranded contaminants, and innate immune sensors can all trigger responses that blunt translation. There are also practical issues like stability during storage and transport, along with degradation over time.

Engineered circular RNA, or circRNA, is being explored as a way to address some of these constraints. Because it lacks free ends, circRNA is naturally more resistant to exonuclease degradation, which may support longer lasting expression. Instead of relying on a 5 prime cap, circRNA constructs can be designed with cap independent initiation elements to drive protein production. In vaccine and therapeutic development, circRNA is increasingly viewed not as a replacement for mRNA, but as a refinement of the format that could expand what the platform can do. This story is not only about engineering. Advances in transcriptome wide sequencing and improved computational tools for detecting backsplice junctions revealed that circRNAs are widespread across tissues and developmental stages. In some cases, they are highly expressed and evolutionarily conserved. That pattern points to regulated biogenesis rather than random splicing noise. Taken together, circRNAs sit at an interesting intersection. They are grounded in basic RNA biology, yet they are steadily moving toward the center of nucleic acid therapeutics.

What is Circular RNA?

A circular RNA is exactly what it sounds like: an RNA molecule that forms a closed loop because its 3′ and 5′ ends are joined together. In eukaryotic cells, most circRNAs are created through a process called back splicing. Instead of being spliced in the usual linear order, a downstream splice donor connects to an upstream splice acceptor, creating a head to tail junction. Because this structure has no free 5′ cap or 3′ poly(A) tail, circular RNA is much less exposed to exonucleases that normally chew up linear RNA from the ends. As a result, circRNAs are often more stable than linear RNAs made from the same sequence.

Biogenesis pathways (endogenous)

CircRNAs can be classified in several ways, but a common functional grouping is:

1. Exonic circRNAs (EcircRNAs): composed of one or more exons, generally exported to the cytoplasm.
2. Exon–intron circRNAs (EIciRNAs): retain intronic sequences and can have nuclear functions.
3. Circular intronic RNAs (ciRNAs): produced from intron lariats that escape debranching and become circular.

Back splicing happens more easily when parts of the RNA are physically brought close together. Certain sequence features help with this. For example, inverted repeat sequences can base pair with each other, causing the RNA to fold so that two splice sites sit next to one another. RNA binding proteins can also act like bridges, holding different parts of the RNA in place and encouraging the loop to form.

Large scale studies in human fibroblasts showed that the regions around circular RNAs often contain matching Alu repeat sequences that can pair up and pull the RNA into the right shape for back splicing. Researchers were also able to systematically identify circular RNAs by treating samples with an enzyme called RNase R, which digests most linear RNAs but leaves circular RNAs intact, and then sequencing what remains. This approach made it possible to detect back splice junctions across the genome.

Most circular RNAs that cells naturally make do not code for proteins. In other words, they are not used as templates to build proteins the way typical messenger RNAs are. However, some circRNAs can be translated into protein. They just do it in a different way. Normal mRNA usually needs a structure at its front end called a 5′ cap to recruit the cell’s protein-making machinery. Circular RNAs do not have that cap, so they cannot use the standard pathway. Instead, certain circRNAs contain special internal sequences, such as internal ribosome entry sites or other nontraditional initiation elements, that allow ribosomes to bind directly to the RNA and start making protein. This is called cap independent translation. It is an alternative way to initiate protein production without relying on the usual cap structure. Modern reviews emphasize that understanding circRNA translation mechanisms is central to engineering circRNA medicines.

How it was discovered

Milestones timeline (selected)

• 1970s: Circular RNAs were identified in pathogens such as viroids (plant pathogens), establishing that circular RNA can exist as a stable biological entity. A landmark 2013 Nature circRNA paper cites the classic viroid work as an early reference point in the field’s history.
• Early 1990s: Reports of “scrambled exons” and circular transcripts appeared in eukaryotic genes, including the testis-specific Sry transcript. Large-scale circRNA mapping later cited these as early demonstrations that exonic circularization occurs in mammals.
• 2012–2013 (RNA-seq era): Multiple groups used high-throughput sequencing and computational methods to show circRNAs are widespread.
• Salzman et al. (2012, PLoS ONE) reported many circular isoforms and argued circRNAs can be the predominant transcript isoform for some genes.
• Jeck et al. (2013, RNA) developed RNase R enrichment (“CircleSeq”) and identified >25,000 backsplice-containing RNA species in human fibroblasts, estimating circRNAs from ~14.4% of actively transcribed genes in that system and highlighting enrichment of flanking complementary Alu repeats.
• Memczak et al. (2013, Nature) detected thousands of circRNAs and provided functional evidence around CDR1as (CDR1 antisense) as a miRNA-binding circRNA with many conserved miR-7 sites and biological effects in vivo.
• Hansen et al. (2013, Nature) provided one of the first clear functional demonstrations of an endogenous circRNA acting as a miRNA sponge (ciRS-7 for miR-7) and also noted Sry as a miR-138 sponge.

Why discovery accelerated

Two developments mattered most:

1. Better detection: Backsplice junctions are not captured by traditional linear transcript assumptions; RNA-seq plus specialized mapping revealed them at scale.
2. Better enrichment/validation: Exonuclease digestion (RNase R) preferentially removes linear RNA, enriching circular forms for discovery and validation.

Two points stand out from circRNA-focused publications:

• Stability & storage: circRNA vaccines are presented as potentially improving on linear mRNA’s instability and innate immunogenicity constraints, at least in principle and in early studies.
• Similarity to mRNA workflow: because both mRNA and coding circRNA ultimately rely on cytosolic translation, many delivery lessons from mRNA (especially LNPs) transfer directly, even if circRNA has distinct manufacturing/purification steps.

What is it being studied as a therapeutic modality?

Engineered circRNA is being explored as a programmable therapeutic platform. The main modalities can be grouped by what the circRNA “does” inside cells:

1. Protein expression (vaccines, protein replacement, cytokines, antibodies)
2. Immune-cell programming in vivo (e.g., transient CAR expression via LNP-delivered RNA)
3. RNA editing / gene regulation tools
4. Immunomodulation via innate immune pathways
5. Diagnostics/biomarkers (endogenous circRNAs)

The science behind key research

A. CircRNAs as endogenous regulators

1. miRNA sponges: Hansen et al. (Nature, 2013) showed the circRNA ciRS-7 contains >70 conserved miR-7 sites, is associated with Argonaute proteins, and can suppress miR-7 activity — supporting the “circRNA as sponge” model; the same paper reported Sry acting as a miR-138 sponge; Memczak et al. (Nature, 2013) similarly highlighted CDR1as/CDR1 antisense as densely bound by miRNA effector complexes with many conserved miR-7 sites, with functional evidence in vivo.
2. Transcriptome-scale abundance and regulated biogenesis: Jeck et al. (RNA, 2013) demonstrated widespread circRNA production and linked circRNA formation to complementary intronic Alu repeats, supporting models where intron pairing facilitates back-splicing.
3. Stability and dynamics: Enuka et al. (Nucleic Acids Research, 2016) reported circRNAs are relatively long-lived and show minimal early changes under certain growth factor stimulation conditions, supporting the idea that circRNAs can be stable cellular components.

B. Innate immune sensing and immunogenicity control

1. Intron identity and “self/non-self” recognition: Chen et al. (Molecular Cell, 2017) reported that transfection of purified, in vitro generated circRNA can induce innate immunity genes and that “self introns” can confer a self-identity signal distinguishing endogenous from foreign circRNA.
2. CircRNAs as regulators of PKR activation: Liu et al. (Cell, 2019) reported endogenous circRNAs tend to form short imperfect duplex regions (16–26 bp) and can act as inhibitors of PKR, with global circRNA degradation (e.g., via RNase L) enabling PKR activation during innate immune responses.

These findings matter directly for therapy: engineered circRNA must often balance enough immune invisibility for translation/durability, versus intentional immunostimulation (vaccines).

C. Engineered circRNA for protein expression

A central “technology leap” was showing that synthetic circRNA can be engineered for robust translation.

1. Wesselhoeft et al. (Nature Communications, 2018) demonstrated engineering strategies enabling potent and stable translation from circular RNA in eukaryotic cells, helping establish circRNA as a practical protein-expression modality rather than only a regulatory ncRNA class.
2. Mechanistic design guidance: A 2024 review synthesizes how circRNA translation can be enabled through non-canonical initiation pathways, emphasizing the roles of intronic pairing, RBPs, and initiation elements in both endogenous and engineered circRNA translation.

D. CircRNA vaccines

A 2023 circRNA vaccine review argues that circRNA vaccines aim to address linear mRNA vaccine constraints (instability, inefficiency, innate immunogenicity) by using circRNA designs that incorporate an IRES and an ORF, with early evidence in models such as SARS-CoV-2 and melanoma.

1. Cold-chain and formulation innovations (lyophilization + targeting): An mBio study (2024) describes a lymph node–targeting circRNA vaccine approach and explicitly connects circRNA to efforts to improve stability and immunogenicity, including after lyophilization, in preclinical models.

E. circRNA in RNA editing and programmable regulation

Beyond protein expression, circRNA can serve as a scaffold for recruiting enzymatic machinery.

1. Circular guide RNAs for ADAR editing (cadRNA): Two Nature Biotechnology papers (2022) describe engineered circular RNAs designed to recruit endogenous ADAR for programmable A-to-I RNA editing, illustrating circRNA as a platform for durable RNA-based editing tools rather than only protein coding.

F. Endogenous circRNAs and anti-tumor immunity

A 2024 Nature paper reported that tumor circRNAs can contribute to anti-tumor immunity through encoded cryptic peptides (linking circRNA translation/antigenicity to immunotherapy concepts).

Current trends

1. Translational focus: stability, redosing, and duration of expression

Preclinical work increasingly emphasizes:

• Longer-lasting antigen/protein expression (relative to typical linear mRNA kinetics)
• Thermostability / storage improvements (including lyophilization)
• Redosable non-viral regimens, especially for immune modulation and cell programming

2. Targeted delivery beyond the liver

RNA therapeutics historically skew toward liver delivery (via GalNAc or certain LNP biodistribution). Current circRNA development places strong emphasis on tissue-targeted nanoparticles and immunotropic delivery — especially for oncology and immune-cell engineering applications.

3. Industry consolidation and investment signals

A major recent signal of “platform validation” is large-pharma activity around engineered circular RNA approaches.

• Eli Lilly–Orna Therapeutics (Feb 2026): Lilly announced an agreement to acquire Orna, describing Orna’s platform as engineered circular RNA plus novel LNPs to enable the body to generate cell therapies (including a CD19-targeting in vivo CAR-T described as “clinical trial-ready”). Orna’s own materials describe its engineered circular RNA (“oRNA”) platform and LNP delivery approach.
• Merck–Orna collaboration (2022): Merck publicly described collaborating with Orna to advance next-generation RNA technology, showing interest from major pharma in circRNA-based platforms even prior to the Lilly deal.
• Parallel “circular-like” industry efforts also exist (e.g., Flagship’s “Endless RNA/eRNA” concept announced publicly), reflecting broader investor interest in long-duration RNA expression platforms.