Small circular RNAs as vaccines for cancer immunotherapy

Zhang, Yu; Liu, Xiang; Shen, Tingting; Wang, Qiyan; Zhou, Shurong; Yang, Suling; Liao, Shimiao; Su, Ting; Mei, Lei; Zhang, Bei; Huynh, Khoa; Xie, Linying; Guo, Youzhong; Guo, Chunqing; Tyc, Katarzyna M.; Qu, Xufeng; Wang, Xiang-Yang; Liu, Jinze; Zhu, Guizhi

doi:10.1038/s41551-025-01344-5

Article
Published: 07 February 2025

Small circular RNAs as vaccines for cancer immunotherapy

Yu Zhang^1,2^na1,
Xiang Liu¹^na1,
Tingting Shen²,
Qiyan Wang ORCID: orcid.org/0000-0001-6379-7879¹,
Shurong Zhou¹,
Suling Yang¹,
Shimiao Liao¹,
Ting Su³,
Lei Mei¹,
Bei Zhang⁴,
Khoa Huynh⁴,
Linying Xie⁵,
Youzhong Guo ORCID: orcid.org/0000-0001-7177-9603⁶,
Chunqing Guo⁷,
Katarzyna M. Tyc⁴,
Xufeng Qu⁴,
Xiang-Yang Wang⁷,
Jinze Liu⁴ &
…
Guizhi Zhu ORCID: orcid.org/0000-0002-2945-7982^1,8

Nature Biomedical Engineering volume 9, pages 249–267 (2025)Cite this article

9496 Accesses
42 Citations
10 Altmetric
Metrics details

Subjects

An Author Correction to this article was published on 12 August 2025

This article has been updated

Abstract

Messenger RNA vaccines have shown strong prophylactic efficacy against viral infections. Here we show that antigen-encoding small circular RNAs (circRNAs) loaded in lipid nanoparticles elicit potent and durable T cell responses for robust tumour immunotherapy after subcutaneous injection in mice, particularly when combined with immune checkpoint inhibition. The small circRNA vaccines are highly stable and show low levels of activation of protein kinase R as well as low cytotoxicity, enabling long-lasting antigen translation (longer than 1 week in cells). Relative to large protein-encoding unmodified or modified mRNAs and circRNAs, small circRNA vaccines elicited up to 10-fold antigen-specific T cells in mice and accounted for 30–75% of the total peripheral CD8⁺ T cells over 6 months. Small circRNA vaccines encoding tumour-associated antigens, neoantigens and oncoviral or viral antigens elicited substantial CD8⁺ and CD4⁺ T cell responses in young adult mice and in immunosenescent aged mice. Combined with immune checkpoint inhibition, monovalent and multivalent circRNA vaccines reduced tumour-induced immunosuppression and inhibited poorly immunogenic mouse tumours, including melanoma resistant to immune checkpoint blockade.

You have full access to this article via your institution.

Download PDF

The role of circular RNAs in regulating resistance to cancer immunotherapy: mechanisms and implications

Article Open access 02 May 2024

Regulation of the tumor immune microenvironment by cancer-derived circular RNAs

Article Open access 16 February 2023

Tumour circular RNAs elicit anti-tumour immunity by encoding cryptic peptides

Article 13 December 2023

Main

Conventional pathogen-based vaccines are associated with virulent reversion, genomic integration, limited biostability, anti-vector immunity, poor pharmacokinetics, weak immunogenicity, and limited benefits for cancer and chronic or pre-existing infections^1,2. mRNA vaccines bypass some of these concerns², as exemplified by the COVID-19 mRNA vaccines, which underscores their potential for versatile applications in disease prevention and therapy^3,4. mRNA vaccines elicit innate immunity by the activation of pattern-recognition receptors (PRRs)⁵ and translate genuine antigens that elicit humoral responses or, upon proteolytic processing for optimal antigen presentation by major histocompatibility complex (MHC), elicit T cell responses. Nevertheless, there is still room to improve mRNA vaccines. For example, despite extensive structural and chemical modifications, current mRNA vaccines have (1) suboptimal biostability that limits the efficiency and duration of antigen translation and immunomodulation; (2) limited thermostability that causes short vaccine shelf-life and demands cold-chain storage and transportation; (3) heterogeneous mRNA impurities (such as double-stranded RNA (dsRNA), N – m, N + m), which are generated during manufacturing by in vitro transcription (IVT) and are difficult to completely remove during purification⁶; and (4) relatively long mRNA and mRNA impurities that may activate protein kinase R (PKR) and 2′-5′-oligoadenylate synthetase and elicit overly strong innate immunity, and thus limit antigen translation efficiency, compromise prophylactic or therapeutic efficacies, and contribute to immunotoxicity^7,8. These challenges also largely apply to self-amplifying mRNA that has long RNA replicons⁹. Furthermore, RNA replicons in self-amplifying mRNA can be sensitive to modifications and may interact with host factors or elicit anti-polymerase immunity that could neutralize future self-amplifying mRNA immunizations¹⁰. circRNA is an emerging class of RNA with unique biological functions^{11,12,13,14,15} and great potential for biomedical application. Natural circRNA is often generated from precursor mRNA back-splicing¹⁶. Recent attempts to develop synthetic circRNA therapeutics and vaccines^{17,18,19,20,21} rely on long mRNA associated with many of the above limitations.

Here we report highly stable small circRNA vaccines that durably produce antigens, and thereby elicit robust and durable T cell responses with great safety for potent tumour combination immunotherapy (Fig. 1). Small circRNAs, typically less than 300 nucleotides, were designed to encode one or many peptide antigens and were synthesized by site-specifically ligating RNA oligonucleotides. An intrinsically fluorogenic aptamer-incorporated small circRNA was also synthesized to study circRNA stability. Small circRNA showed high thermostability (solution storage half-life of around 400 days at −20 °C) and biostability than not only the corresponding small linearized RNA but also state-of-the-art modified mRNA and large circRNA (typically larger than 1,000 nucleotides); furthermore, circRNA sequencing (circRNA-seq) verified the nucleotide authenticity of small circRNA transfected into live cells. The high stability of small circRNA enabled sustained peptide expression for at least a week in cells. Small circRNA vaccines activated the retinoic acid-inducible gene I (RIG-I) immunostimulation pathway in dendritic cells (DCs). Furthermore, compared with large nucleoside-modified mRNA and large circRNA, small circRNA vaccines showed good safety and low undesired PKR activation. Small circRNAs were efficiently loaded into nanocarriers, such as ionizable lipid nanoparticles (LNPs), for efficient delivery to antigen-presenting cells (APCs) in draining lymph nodes. Coupled with the innate immunostimulation by circRNA LNPs, the sustained antigen translation by small circRNA over a prolonged duration enables efficient and persistent antigen presentation and T cell priming. As a result, small circRNA vaccines encoding MHC-II- and MHC-I-restricted antigens elicited potent and durable (with memory) CD4⁺ and CD8⁺ T cell responses, respectively, both of which are pivotal for cancer therapy. Relative to multiple types of peptide/protein-encoding modified mRNA, unmodified mRNA and large circRNA, modification-free small circRNA vaccine elicited stronger and more long-lasting T cell responses that lasted at least 6 months post immunization. This makes small circRNA vaccines promising for in vivo generation of therapeutically significant doses of T cell receptor (TCR)-T cells, without any ex vivo T cell engineering and manufacturing. Moreover, small circRNA elicited profound T cell responses in immunosenescent aged mice. Modular small circRNA was easily adjusted to elicit T cell responses against various antigens, including tumour-associated antigens, tumour neoantigens, oncoviral antigens and viral antigens. In mice, small circRNA vaccines reduced tumour immunosuppression and enhanced tumour infiltration of antitumour immune cells. Monovalent or multivalent small circRNA vaccines, alone or combined with immune checkpoint blockade (ICB), notably inhibited or eradicated tumours in multiple syngeneic tumour models, including poorly immunogenic and ICB-resistant Braf^V600E melanoma. These results demonstrate the great potential of small circRNA vaccines for tumour combination immunotherapy.

**Fig. 1: Schematic illustration of highly stable small circRNA vaccines that elicit potent and long-lasting T cell responses for tumour immunotherapy.**

Results

Optimization and characterization of highly stable small circRNA vaccines

Small circRNAs consist of minimal elements: codon-optimized peptide-antigen-encoding RNA and a short internal ribosome entry site (IRES) and a Kozak consensus sequence that recruit ribosomes for peptide translation (Fig. 1). Using MHC-I-restricted ovalbumin₂₅₇_–₂₆₄ (OVA₂₅₇_–₂₆₄ or SIINFEKL) as a model antigen, we prepared circRNA as follows: (1) circularizing 5′-phosphate-RNA using T4 RNA ligases and complementary 30-mer DNA splints; (2) removing linear/lariat RNA and DNA using DNase I and exonuclease T or RNase R²²; (3) purification to remove immunostimulatory small DNA/RNA fragments, enzymes and nucleotides (Supplementary Fig. 1); and (4) verification by Sanger sequencing (Fig. 2a) and gel electrophoresis with an estimated circularization rate of 32% (Supplementary Fig. 1). A short IRES that efficiently recruits ribosomes and initiates translation is critical for the optimal immunomodulatory efficacy of small circRNA vaccines²³. As shown by H-2K^b/SIINFEKL staining on vaccine-treated DCs, circRNA-SIINFEKL with 71-mer crucifer-infecting tobacco mosaic virus (crTMV) IRES enabled efficient antigen presentation, relative to circRNA-SIINFEKL with short IRES(es) long interspersed nuclear elements 1 (LINE1) and RNA binding motif protein 3 (RBM3), the corresponding linear RNA (liRNA, linearized circRNA with the nicking site between IRES and coding sequence, without 5′ cap or 3′ poly(A) or untranslated regions (UTRs) (Fig. 2a)), and CpG oligonucleotide-adjuvanted OVA protein^24,25,26 (Fig. 2b,c, Supplementary Fig. 2a and Supplementary Table 1). Consistently, crTMV enabled circRNA-SIINFEKL to efficiently prime SIINFEKL-specific B3Z CD8⁺ T cell hybridoma (Fig. 2b–d). Deletion of IRES or Kozak sequence reduced antigen presentation and T cell priming (Supplementary Fig. 2b). Thus, crTMV IRES was selected for further studies. Interestingly, relative to state-of-the-art 5-methoxyuridine-modified OVA-encoding mRNA (5moU-mRNA-OVA), circRNA-SIINFEKL enhanced and prolonged antigen presentation and T cell activation (Fig. 2b,d and Supplementary Fig. 2b). A model Flag peptide-encoding circRNA (circRNA-Flag) efficiently expressed Flag products in live mouse DC2.4 cells, as shown by anti-Flag immunostaining (Fig. 2e).

**Fig. 2: Highly stable small circRNA vaccines for efficient peptide translation and antigen presentation.**

mRNA can be degraded by ubiquitous biological and environmental RNases, especially exonucleases that are primarily responsible for mRNA degradation¹, which limits mRNA shelf-life, biological half-life, antigen translation efficiency and duration, and consequently immunomodulatory efficacy. The absence of termini in circRNA prevents exonuclease degradation¹⁷. To study the biostability of circRNA in live cells, we synthesized a model circRNA in which the coding RNA was substituted with a fluorogenic RNA aptamer, Broccoli²⁷. Upon transfection, the resulting circBroccoli showed excellent biostability in live DCs and the Broccoli fluorescence sustained for at least 7 days despite signal dilution during cell division, in contrast to almost complete fluorescence decay of linearized circBroccoli within 12 h (Fig. 2f–h and Supplementary Fig. 3a). Exogenous RNA, including circRNA, can be subject to mutation owing to RNA editing by endogenous editors, such as cytidine deaminases AID and APOBEC²⁸, and adenosine deaminase ADAR²⁹. Yet, such circRNA mutation may affect the ability of IRES and Kozak sequence to initiate peptide translation and alter peptide sequences. To study this, mouse bone marrow-derived dendritic cells (BMDCs) were transfected with circRNA-SIINFEKL. Twenty-four hours later, total circRNA was isolated from cells, and circRNA-seq analysis of circRNA-SIINFEKL showed no detectable circRNA mutations (Fig. 2i), suggesting the high sequence authenticity of this small circRNA in live cells. Modification-free small circRNA showed superior thermostability than not only the corresponding small liRNA but also 5moU-mRNA-OVA. Specifically, as estimated with a one-phase decay model, upon storage in phosphate-buffered saline (PBS) at –20 °C, 4 °C and 23 °C, the half-lives of circRNA-SIINFEKL are 401, 78 and 16 days, respectively, in contrast to 143, 44 and 6 days, respectively, for 5moU-mRNA-OVA, and 2.6, <0.5 and <0.5 days, respectively, for liRNA-SIINFEKL (Fig. 2j and Supplementary Fig. 3b). Consistently, when transfected into DCs, small circRNA-Flag showed durable Flag expression for at least 7 days when the experiment ended, in contrast to less than 3 days of protein expression by 5moU-mRNA-fLuc (Fig. 2k). Interestingly, circBroccoli-loaded LNPs (Fig. 2l) retained strong fluorescence intensity upon transfection in DCs after storage in solutions at 4 °C or −20 °C (supplemented with sucrose) for 70 days, verifying great thermostability of circRNA LNPs (Fig. 2m). circRNA is eventually degraded by hydrolysis³⁰, endonuclease degradation³¹ and exonucleases upon circRNA linearization. Large circRNA has been often used for drug and vaccine development so far¹¹. We hypothesize that because small circRNA typically has fewer sites of hydrolysis and endonuclease cleavage than large circRNA, small circRNA is overall more stable than large circRNA. To test this, we synthesized 1,760-mer large circRNA consisting of a long IRES (CVB3) and RNA encoding SARS-CoV-2 spike protein receptor-binding domain (RBD) (Supplementary Fig. 4). Upon incubation in diluted fetal bovine serum (FBS) followed by gel electrophoresis, 111-mer small circRNA-SIINFEKL showed higher stability than large circRNA-RBD (Fig. 2n,o). Overall, small circRNA showed excellent biostability and thermostability, which are critical for their long shelf-life, sustained antigen translation, and robust and durable immunomodulation.

LNPs efficiently delivered small circRNA to lymph nodes and APCs to elicit T cell responses

Nanocarriers are pivotal for the delivery and immunomodulation of mRNA vaccines³². To select a nanocarrier for small circRNA vaccines, we tested liposomes and three LNPs (with ionizable lipids SM-102, Dlin-MC3-DMA and Dlin-KC2-DMA, respectively), two classes of the most successful nucleic acid nanocarriers thus far³². We loaded circRNA-SIINFEKL and control 5moU-mRNA-OVA into these nanocarriers (N/P ratio 6). SM-102 LNPs showed comparable encapsulation efficiencies for mRNA (86.3%) and circRNA (86.9%). Owing to the small size of small circRNA, 185 copies of circRNA-SIINFEKL were calculated to be loaded per LNP with 170 nm in diameters, in contrast to 14 copies of 5moU-mRNA-OVA per LNP. Upon subcutaneous injection at the tail base of C57BL/6 mice, H-2K^b-SIINFEKL tetramer staining showed that SM-102 LNPs of 5 μg circRNA-SIINFEKL (10–15% typical doses of T cell mRNA vaccines³³) elicited the most frequent SIINFEKL⁺CD8⁺ T cells among peripheral blood mononuclear cells (PBMCs) (Fig. 3a,b). Furthermore, as shown by Luminex, among all formulations, SM-102 LNPs of circRNA-SIINFEKL elicited the lowest levels of systemic chemokines associated with reactogenicity, a common side effect of mRNA vaccines (Fig. 3c and Supplementary Fig. 5). This indicates the good safety potential of small circRNA vaccines. Taken together, SM-102 LNP was selected as the nanocarriers for small circRNA vaccines. In vivo imaging system (IVIS) imaging showed that, for at least 8 days after subcutaneous injection in mice, SM-102 LNPs enhanced the accumulation of IR800-labelled circRNA (IR800-circRNA) in draining lymph nodes, which harbour various lymphocytes and orchestrate immunomodulation (Fig. 3d–f). This provides the basis for sustained antigen expression and T cell priming over a prolonged duration. Lymph nodes harbour various APC subsets that have differential vaccine uptake and immunostimulation efficacies. We studied circRNA uptake by intranodal APCs by flow cytometry 24 h after subcutaneous injection of Cy5-circRNA LNPs in mice. LNPs efficiently delivered circRNA to intranodal APC subsets that are pivotal to present peptide epitopes and elicit antigen-specific T cell responses, including CD8⁺ T cell-priming CD11c⁺CD11b⁻CD8⁺ classical DCs (cDCs), CD4⁺ T cell-priming CD11c⁺CD11b⁺CD4⁺CD103⁻CD205⁺ cDCs and CD11c⁺CD11b⁻CD8⁻CD4⁻CD103⁺ migratory cDCs (Fig. 3g). Moreover, within only 0.5 h after transfection in cultured DCs, LNPs enabled rapid endosome escape of circRNA to the cytosol to allow peptide translation (Fig. 3h,i and Supplementary Fig. 6). It is worth noting that liposomes also efficiently delivered small circRNA to draining lymph nodes and intranodal APCs in mice (Extended Data Figs. 1 and 2) and elicited SIINFEKL-specific CD8⁺ T cell responses more efficiently than 5moU-mRNA-OVA and CpG-adjuvanted OVA (Extended Data Fig. 3a–c). Meanwhile, presumably owing to vaccine-induced immunostimulation, liposomal circRNA-SIINFEKL upregulated expression of the immune checkpoint programmed death receptor 1 (PD-1) on T cells, especially SIINFEKL-specific CD8⁺ T cells (Extended Data Fig. 3d–f), providing an opportunity for ICB combination therapy with circRNA vaccines. Consistently, liposomal circRNA vaccine-immunized mice resisted challenge with EG7.OVA tumour cells, without significant weight loss, indicating the good safety potential of these vaccines (Extended Data Fig. 3g,h). Overall, nanocarriers such as LNPs efficiently delivered small circRNA vaccines to lymph nodes and pivotal intranodal APCs to elicit potent T cell responses with great safety potential.

Small circRNA mediated immunomodulation with low PKR activation

Various types of RNA have been found to elicit innate immunity by activating endosomal PRRs such as Toll-like receptor 3 (TLR3), TLR7, TLR8 or cytosolic PRRs (such as RIG-I and melanoma differentiation-associated protein 5 (MDA5))³⁴. Combined with the immunostimulatory ability of nanocarriers (for example, LNPs), mRNA vaccines elicit innate immunity that provides APCs with pro-inflammatory cytokines and co-stimulation signals for antigen presentation and T cell priming. To investigate the impact of small circRNA vaccines on immunomodulation, we conducted RNA-seq genome-wide transcriptomic analysis of mouse BMDCs LNP-transfected with circRNA-SIINFEKL and 5moU-mRNA-OVA, respectively, for 24 h, with PBS or blank LNPs as controls. About 560 genes were grouped by their involvement in inflammation, migration, antigen uptake and presentation, PRRs (such as RIG-I-like receptors (RLRs), TLRs and C-type lectin receptors (CLRs)), MHC-I and MHC-II, or miscellaneous immune-related genes. From the hierarchical clustering of these differentially expressed genes (Fig. 4a,b), we identified upregulated PRRs and activated immune-related pathways by LNP-transfected circRNA relative to blank LNPs. Overall, relative to blank LNPs, circRNA-SIINFEKL showed low upregulation of the above immune-related genes. For example, circRNA upregulated the transcriptions of Ddx58 (also known as Rigi) and Clec4, whereas 5moU-mRNA-OVA activated RLRs (Ifih1 (also known as Mda5), Oas1b, Eif2s1 and Eif2ak2 (Pkr)) and Tlr3 (Fig. 4c) that are characteristic of long dsRNA/ssRNA PRR activation. KEGG pathway analysis further confirmed that RNA sensor genes such as Rigi were activated by circRNA-SIINFEKL (Supplementary Figs. 7–10). The differential PRR activation profiles between circRNA-SIINFEKL and 5moU-mRNA-OVA may be due to structural differences such as the absence of long dsRNA in small circRNA. Furthermore, circRNA-SIINFEKL matured DCs, as indicated by enhanced gene activities of MHC-I (H2-K1) and co-stimulatory molecules (such as Cd40, Cd80 and Cd86) (Fig. 4c), which was supported by the elevated expression levels of co-stimulatory factors CD80 and CD86 (Supplementary Fig. 11). Moreover, circRNA-SIINFEKL triggered a multifaceted shift to inflammation in DCs, with upregulation of NLR (Nlrc4) and pro-inflammatory cytokines (Tnf and Il12rb1). circRNA-SIINFEKL also upregulated some immunosuppressive mediators, which presumably act as endogenous negative regulators of innate immune activation. Some of these genes encode therapeutic targets, for example, programmed death ligand 1 (PD-L1; encoded by Cd274), that hold potential for combination therapy with circRNA vaccines. To validate PRRs for circRNA, HEK-Lucia RIG-I, HEK-Blue hTLR3 and HEK-Blue hTLR7 reporter cells were respectively transfected with circRNA-SIINFEKL with controls, including large circRNA-OVA, 5moU-mRNA-OVA and blank transfection agent, as well as RNA immunostimulant polyinosinic-polycytidylic acid (poly(I:C)) as a positive control. circRNA-OVA comprises IRES CVB3 and a coding sequence (CDS) for OVA, and was synthesized by IVT and permuted intron–exon I (PIE-I)-mediated RNA circularization (Supplementary Table 2 and Supplementary Fig. 12a,b). As a result, circRNA-SIINFEKL activated RIG-I, but not TLR3 or TLR7 (Fig. 4d). RIG-I-dependent circRNA sensing was further validated in model RM1 cells with knockout of interferon-β (IFNβ) promoter stimulator I (IPS-1), an adaptor required by RIG-I-mediated type I IFN induction (Fig. 4e). These results provide the basis for future full understanding of the immunomodulation mechanism by small circRNA vaccines. Lastly, relative to large circRNA-RBD and 5moU-mRNA-OVA, small circRNA-SIINFEKL showed low PKR phosphorylation (pPKR) in cells, indicating low PKR activation (Fig. 4f,g and Supplementary Fig. 12c,d). This suggests that, relative to large circRNA and mRNA often with long dsRNA, the minimal long dsRNA in small circRNA minimizes pPKR inhibition of protein translation. Consistently, relative to large circRNA-RBD, small circRNA-SIINFEKL resulted in low cytotoxicity in HEK293T cells (Fig. 4h). Overall, these results reveal the immunomodulatory mechanism of small circRNA vaccines with low PKR activation and low cytotoxicity.

**Fig. 4: Small circRNA vaccine activated PRRs for intrinsic immunostimulation with low PKR activation.**

Benchmarking small circRNA vaccines against modified mRNA, large circRNA and unmodified mRNA vaccines

Current mRNA vaccines use long poly(A) and artificial caps, as well as modified nucleotides for modified mRNA, which together result in balanced mRNA biostability, protein production and immunostimulation. Moreover, synthetic large circRNAs have been engineered as a novel class of mRNA vaccines against SARS-CoV-2 (ref. ¹⁷). We first studied the immunomodulatory efficacies of modification-free circRNA-SIINFEKL benchmarked against three forms of modified mRNA vaccines in vitro and in mice: (1) protein-encoding 5moU-mRNA-OVA, (2) SIINFEKL-encoding 5moU-mRNA (5moU-mRNA-SIINFEKL) and (3) pseudouridine (ψ)-mRNA-SIINFEKL. All these mRNAs have 5′/3′ UTRs, 3′ A120 and CleanCap (Supplementary Tables 1 and 2). After LNP transfection into DC2.4 cells, circRNA-SIINFEKL elicited the most rapid, most abundant and most durable SIINFEKL antigen presentation over 3 days (Fig. 5a and Supplementary Fig. 13). Next, C57BL/6 mice (6–8 weeks) were immunized with these RNAs at escalating doses (1, 3, 10, 30 and 100 μg), respectively, followed by monitoring their potency and duration of SIINFEKL-specific CD8⁺ T cell responses over 180 days, body weights and innate immune responses to evaluate T cell responses and vaccine safety (Fig. 5b). After two doses, tetramer staining (day 21) showed that while all these vaccines showed dose-dependent SIINFEKL⁺CD8⁺ T cell responses, circRNA-SIINFEKL elicited the largest fractions of SIINFEKL⁺CD8⁺ T cells among all PBMC CD8⁺ T cells (Fig. 5c and Supplementary Fig. 14). As low as 2 × 1 μg circRNA-SIINFEKL elicited 2.6- and 7.3-fold PBMC SIINFEKL⁺CD8⁺ T cell fractions relative to that by 5moU-mRNA-SIINFEKL and ψ-mRNA-SIINFEKL, respectively. Furthermore, at higher doses, circRNA-SIINFEKL caused less mouse body weight loss and more rapid weight recovery than benchmark mRNAs, indicating the potentially superior safety of small circRNA vaccines (Fig. 5d and Supplementary Fig. 15). A second booster of circRNA-SIINFEKL (day 28) substantially expanded the repertoire of SIINFEKL-specific CD8⁺ T cells, which plateaued at ~75% peripheral SIINFEKL-specific PBMC CD8⁺ T cells (day 35) at a dose of 3 × 30 μg (Supplementary Fig. 16). Over days 21–180, while all RNA vaccines showed dose-dependent SIINFEKL-specific T cell responses, the area under the curve (AUC) of SIINFEKL-specific PBMC CD8⁺ T cell fractions elicited by circRNA-SIINFEKL was higher than that by all the benchmark mRNA vaccines (Fig. 5e). Moreover, on day 35, the total counts of SIINFEKL⁺CD8⁺ T cells in peripheral blood and spleen per mouse induced by as low as 10 μg circRNA-SIINFEKL were estimated to be over 2-fold typical dose of adoptive TCR-T cells used for mouse solid tumour immunotherapy³⁵ (Fig. 5f). This suggests the potential of small circRNA vaccines for in vivo generation of therapeutically significant dose of TCR-T cells for tumour immunotherapy, which bypasses any ex vivo TCR-T cell engineering and manufacturing. The T cell responses induced by circRNA-SIINFEKL waned less rapidly than that by modified mRNA, although high-dose 5moU-mRNA-OVA, which produces more immunogenic OVA than SIINFEKL, also elicited up to ~70% SIINFEKL-specific PBMC CD8⁺ T cells (day 35). Specifically, on day 180, 8–30% SIINFEKL-specific PBMC CD8⁺ T cells persisted with 10–100 μg circRNA-SIINFEKL, whereas such T cell fractions elicited by modified mRNA-SIINFEKL waned almost completely and that by 10-100 μg 5moU-mRNA-OVA waned to 3–20%. This suggests that circRNA-SIINFEKL elicited superior T cell memory than these benchmark mRNAs. The immune memory was verified by elevated total fractions of CD62L^–CD44^hi effector memory T (T_EM) cells and D62L⁺CD44^hi central memory T (T_CM) cells, especially SIINFEKL⁺ memory T cells (Supplementary Fig. 17). Consistently, although the total counts of PBMC and splenic SIINFEKL⁺CD8⁺ T cells elicited by all vaccines reduced on day 180 relative to day 35, the T cell repertoires elicited by circRNA vaccines persisted for a longer duration than those elicited by benchmark mRNA (Fig. 5f). We expect that additional booster circRNA immunization would maintain the high T cell repertoires, when needed. Immunostimulation is often accompanied by immune exhaustion, as characterized by the upregulated expression of immune checkpoints such as PD-1. Levels of PD-1 in total PBMC CD8⁺ T cells in the immunized mice suggested dose-dependent PD-1 upregulation by day 35 (Supplementary Fig. 18), and PD-1 levels had largely recovered by day 49. The immune checkpoint upregulation can sensitize T cells for ICB, providing an opportunity to combine ICB with circRNA vaccines for optimal cancer therapeutic efficacy. Next, we benchmarked circRNA-SIINFEKL against unmodified OVA-encoding mRNA-OVA and large circRNA-OVA vaccines for their abilities to elicit T cell responses in mice. circRNA-OVA consists of an efficient long IRES (CVB3) and encodes highly immunogenic full-protein OVA (Supplementary Fig. 12a,b). Upon immunization in C57BL/6 mice (6–8 weeks), circRNA-SIINFEKL elicited higher fractions of PBMC SIINFEKL⁺CD8⁺ T cells than unmodified mRNA-OVA and large circRNA-OVA at equivalent doses over 120 days (Fig. 5g and Extended Data Fig. 4).

**Fig. 5: Small circRNA vaccine outperformed several benchmark modified mRNA and large circRNA vaccines to elicit robust and durable T cell responses with great safety in young adult mice and aged mice.**

To study vaccine-elicited innate immunity associated with reactogenicity and pro-inflammation, we measured a panel of serum chemokines and pro-inflammatory cytokines by Luminex 12 h after the third dosing. Overall, relative to modified mRNAs and large circRNA-OVA, small circRNA-SIINFEKL induced lower levels of chemokines and cytokines that are often associated with immunotoxicity (Fig. 5h and Supplementary Fig. 19). This suggests the good safety potential of small circRNA vaccines. Overall, these results demonstrate the superior or at least comparable T cell responses and safety of small circRNA vaccines relative to modified mRNA and large circRNA vaccines.

Immunosenescence during ageing makes older people vulnerable to immune disorders and impairs vaccines’ immunomodulatory efficacy³⁶. Thus, vaccines are highly desired to elicit potent and durable immunity in older people. In aged C57BL/6 mice (1 year old) (Fig. 5i,j), circRNA-SIINFEKL (5 μg, days 0 and 14) elicited 184% more frequent PBMC SIINFEKL⁺CD8⁺ T cells than 5moU-mRNA-OVA and 242% more than CpG-adjuvanted OVA (day 21), with potent T cell polyfunctionality and immune memory that protected mice from EG7.OVA cell challenge (Fig. 5i–k and Supplementary Fig. 20). These results suggest the ability of small circRNA vaccines to potentiate the T cell responses with memory in immunosenescent aged mice.

Broad application of circRNA vaccines for various peptide immunogens

Modular small circRNA vaccines can be easily adjusted to encode versatile peptide immunogens for broad applications. Beside MHC-I-restricted immunogens, MHC-II-restricted immunogens elicit CD4⁺ helper and effector T cell responses that are crucial for disease prophylaxis and immunotherapy. In C57BL/6 mice, circRNA encoding MHC-II-restricted OVA₃₂₃_–₃₃₉ (named as ISQ) (5 μg; days 0 and 14) elicited ISQ-specific polyfunctional CD4⁺ T cells, as shown by intracellular IFNγ/TNF (formerly known as TNFα) staining (day 21) (Fig. 6a,b). Interestingly, circRNA-ISQ enhanced the ability of MHC-I-restricted circRNA-SIINFEKL to elicit SIINFEKL⁺CD8⁺ T cell response, likely because ISQ⁺CD4⁺ helper T cell response promoted SIINFEKL⁺CD8⁺ T cell priming³⁷ (Fig. 6c). Moreover, circRNA-ISQ + circRNA-SIINFEKL outperformed 5moU-mRNA-OVA, which encodes OVA protein with both SIINFEKL and ISQ epitopes, to elicit PBMC SIINFEKL⁺CD8⁺ T cell response. Furthermore, small circRNA vaccines are applicable to tumour-associated antigens and tumour neoantigens for cancer immunotherapy. For example, circRNA encoding MHC-I-restricted ADPGK, an MC38 tumour cell-specific neoantigen³⁸, elicited dose-dependent T cell responses with memory in mice (Fig. 6d,e and Supplementary Fig. 21). circRNA-ADPGK (2 × 5 μg) elicited 5-fold ADPGK ⁺CD8⁺ T cells, relative to liRNA-ADPGK or CpG-adjuvanted ADPGK peptide; meanwhile, circRNA-ADPGK elevated the PD-1 level on CD8⁺ T cells, which can sensitize these cells for anti-PD-1 ICB in cancer combination immunotherapy (Fig. 6f). Multivalent vaccines can overcome the challenges of tumour antigenic heterogeneity and tumour cell immune evasion³⁹. Bivalent circRNA encoding MHC-I-restricted melanoma-associated antigens mouse TRP2₁₈₀_–₁₉₀ and human gp100₂₃_–₃₃, the latter priming T cells to recognize both human and mouse gp100 (ref. ⁴⁰), elicited bispecific CD8⁺ T cells (Fig. 6g). Lastly, T cell vaccines hold great potential to prevent and treat pathogenic viral infections and associated cancers, such as SARS-CoV-2 and human papillomavirus (HPV) infections, and HPV-associated cancers⁴¹. Relative to long B cell epitopes that are often susceptible to mutations, T cell peptide epitopes are short (8–20 amino acids) and highly conserved owing to less susceptibility to mutations. This makes T cell epitopes attractive to elicit broadly responsive T cell responses against rapidly mutating pathogens, such as SARS-CoV-2. As a step to this end, circRNA encoding MHC-I-restricted HPV16 E7₄₉_–₅₇ (ref. ⁴¹) outperformed liRNA-E7₄₉_–₅₇ and CpG-adjuvanted E7₄₉_–₅₇ peptides to elicit potent and long-lasting E7₄₉_–₅₇-specific CD8⁺ T cell response in mice (Fig. 6h,i and Supplementary Fig. 22). Likewise, circRNA encoding MHC-I-restricted SARS-CoV-2 spike protein RBD₄₄₀_–₄₅₉ elicited potent and long-lasting RBD₄₄₀_–₄₅₉⁺CD8⁺ T cells, as shown by tetramer staining over 70 days and intracellular cytokine staining (Fig. 6j,k). Overall, these results demonstrate the broad applicability of small circRNA vaccines to elicit T- cell responses.

**Fig. 6: Broad application of small circRNA vaccines to elicit T cell responses against various peptide antigens.**

Low-dose circRNA vaccines reduced tumour immunosuppression for potent immunotherapy

Tumour microenvironment (TME) is primarily where tumour cells suppress antitumour immunity². In MC38 tumours, MC38-specific neoantigen vaccine circRNA-ADPGK, especially when combined with anti-PD-1, reduced immunosuppressive myeloid-derived suppressor cells (MDSCs) and CD4⁺FOXP3⁺CD25⁺ regulatory T (T_reg) cells, increased the tumour infiltration of total CD8⁺ T cells and ADPGK⁺CD8⁺ T cells, and the ratio of enhanced CD8⁺ T cells to T_reg cells in the tumour, which predicts tumour therapeutic efficacy (Fig. 7a–d). Consistently, RNA-seq of tumour tissue and transcriptome analysis revealed that circRNA-ADPGK and circRNA-ADPGK + anti-PD-1 triggered a multifaceted shift of the tumour to an inflamed and tumoricidal microenvironment, with upregulated gene transcripts related to pro-inflammatory cytokines, leukocyte-recruiting chemokines, factors associated with DC maturation and T cell priming, and natural killer (NK) cell and macrophage activation (Fig. 7e–g). These changes of the cellular and molecular tumour immune microenvironment are expected to be important attributes to the tumour therapeutic efficacy. As a result, circRNA-ADPGK neoantigen vaccine (3 × 5 μg) inhibited the growth of ADPGK-positive MC38 tumour (Fig. 7h). Moreover, circRNA-ADPGK potentiated the therapeutic efficacy of anti-PD-1, which blocks the upregulated PD-1 on energic T cells (Fig. 7h). Depletion of CD8⁺ T cells, but not CD4⁺ T cells or NK cells, abrogated the therapeutic efficacy of circRNA, verifying the central role of CD8⁺ T cells in MHC-I-restricted circRNA-ADPGK (Fig. 7i).

**Fig. 7: Small circRNA neoantigen vaccine reduced tumour immunosuppression for potent tumour immunotherapy.**

Anticancer T cell responses and TME immunity are pivotal for effective cancer immunotherapy. Given their ability to elicit potent and durable T cell responses and reduce TME immunosuppression, small circRNA vaccines were further assessed for tumour immunotherapy in multiple mouse tumour models. Low-dose circRNA-SIINFEKL (3 × 5 μg) regressed EG7.OVA tumours (~50 mm³) and prolonged mouse survival, which outperformed benchmark 5moU-mRNA-OVA and CpG-adjuvanted OVA (Extended Data Fig. 4a–d). Depletion of CD8⁺ T cells, but not CD4⁺ T cells or NK cells, abrogated the therapeutic efficacy, verifying the central role of CD8⁺ T cells in MHC-I-restricted circRNA-SIINFEKL. circRNA did not cause changes in mouse body weights, indicating its good safety potential (Extended Data Fig. 4d). Moreover, small circRNA-SIINFEKL outcompeted full-protein OVA-encoding circRNA-OVA with a highly efficient IRES (CVB3) to inhibit EG7.OVA tumour growth in C57BL/6 mice (Extended Data Fig. 4e). High-risk HPV, including HPV 16, causes ~5% of all cancer cases. However, current US Food and Drug Administration (FDA)-approved HPV vaccines are unable to treat HPV-associated cancer. Moreover, current immunotherapy such as ICB and adoptive cell transfer are limited by ICB resistance and poor T cell memory. Therapeutic vaccines are promising to elicit T cell response for combination immunotherapy of HPV-associated cancer. MHC-I-restricted HPV16 protein E7 is an attractive oncoviral antigen because it is consistently and selectively expressed in HPV-associated cancer but not healthy cells, bypasses host immune tolerance, and is required to initiate and maintain HPV-associated cancer, which prevents cancer immune escape. Yet, current E7-based experimental therapeutic vaccines have limited efficacy against HPV-associated cancer in the clinic so far owing to low immunogenicity. Given the robust T cell response elicited by MHC-I-restricted circRNA-E7 (Fig. 6h,i), we tested circRNA-E7 for the immunotherapy of E7-positive TC-1 tumour. Low-dose circRNA-E7 (3 × 5 μg) inhibited TC-1 tumour progression via CD8⁺ T cells (Fig. 8b and Extended Data Fig. 4f). Because circRNA-E7 upregulated PD-1 on CD8⁺ T cells, combining circRNA-E7 with anti-PD-1 further enhanced the tumour therapeutic efficacy (Fig. 8b,c). We expect that increasing the circRNA dose would further improve the tumour therapeutic efficacy.

**Fig. 8: Monovalent or multivalent small circRNA vaccines for robust combination immunotherapy of multiple types of tumour.**

Tumour antigenic heterogeneity and the resulting tumour cell immune escape represent a major obstacle against subunit vaccine-based tumour immunotherapy. Multivalent vaccines can at least partially overcome tumour antigenic heterogeneity and prevent tumour immune escape, thereby promoting tumour therapeutic efficacy. To test this using small circRNA vaccines, we engineered tetravalent small circRNA, termed as circRNA-T2/g/T1/T1, which encoded MHC-I-restricted mouse TRP2₁₈₀_–₁₉₀, human gp100₂₃_–₃₃ and mouse TRP1₄₅₅_–₄₆₃, as well as MHC-II-restricted mouse TRP1₁₀₆_–₁₃₀ (ref. ³⁹). In poorly immunogenic B16F10 melanoma, MHC-I/II-restricted circRNA-T2/g/T1/T1 (3 × 10 μg) inhibited tumour progression, in which both CD8⁺ and CD4⁺ T cells were critical (Fig. 8d,e). Combining circRNA with both anti-PD-1 and anti-CTLA-4 further potentiated the therapeutic efficacy and prolonged mouse survival (Fig. 8d–g), with 2 out of 7 tumours showing complete regression without recurrence in 3 months (Fig. 8f). Around 50% of human melanoma contain BRAF^V600E mutations, which cause constant oncogenic activation of BRAF, a serine/threonine protein kinase, and promotes melanoma tumorigenesis by, for example, evasion of antitumour immune responses, senescence and apoptosis. Syngeneic Braf^V600E SM1 mouse melanoma is highly aggressive and poor immunogenic with highly immunosuppressive TME, making it often resistant to immunotherapy. Thus, SM1 melanoma closely resembles human BRAF^V600E melanoma to predict its clinical therapeutic efficacy. Indeed, in syngeneic C57BL/6 mice, Braf^V600E SM1 mouse melanoma did not respond to even dual ICB (anti-PD-1 + anti-CTLA-4). By contrast, tetravalent circRNA-T2/g/T1/T1 vaccine (3 × 30 μg) inhibited the progression of Braf^V600E SM1 melanoma, and combining circRNA with anti-PD-1 + anti-CTLA-4 completely regressed 4 out of 8 tumours with 50% survival rate in 2 months (Fig. 8h,i). Overall, these results demonstrate the great potential of small circRNA vaccines for tumour combination immunotherapy.

Discussion

mRNA vaccines have been a great success as prophylactic COVID-19 vaccines and have shown promising clinical responses for cancer immunotherapy. Nonetheless, current mRNA vaccines are still associated with suboptimal stability, low loading capacity of very large mRNA in nanocarriers, heterogeneous by-products from IVT (such as truncated or extended mRNA, dsRNA), suboptimal immunomodulatory potency and duration, and reactogenicity. Addressing these issues may advance the development of mRNA vaccines, especially for the therapy of pre-existing diseases, such as cancer. To this end, we developed highly stable modification-free small circRNAs as mRNA vaccines to elicit potent and long-lasting T cell responses for robust tumour immunotherapy with great safety. Small circRNA is synthesized by ligating automatically synthesized RNA oligonucleotides, which could facilitate circRNA manufacturing and bypass purification complications associated with mRNA complex structures and mRNA by-products. Small circRNA has high loading capacity in nanocarriers for efficient delivery into draining lymph nodes and APCs. Small circRNA showed superior thermostability and biostability than not only small liRNA but also modified mRNA and large circRNA, the latter likely because of fewer sites of endonuclease cleavage and hydrolysis per small circRNA molecule than large circRNA. Moreover, circRNA-seq revealed the high nucleotide authenticity of circRNA in DCs, ruling out circRNA mutation under this condition owing to, for example, endogenous RNA editing. Highly stable small circRNA allowed for efficient and long-lasting antigen translation over at least 1 week. crTMV IRES enabled circRNA vaccine for efficient antigen presentation and T cell priming, presumably owing to strong ribosome binding with the structurally intact IRES in circRNA. The mechanism of peptide translation from small circRNA remains to be fully understood. Presumably, without a stop codon following peptide-coding RNA, circRNA is translated to a fused antigen and partial IRES-encoding peptide until reaching the intrinsic stop codons in crTMV IRES. Furthermore, if ribosomes skip discrete intrinsic stop codons⁴² in crTMV IRES, circRNA may produce long concatemeric antigenic peptides via rolling circle translation. Overall, these long peptide antigens would mimic ‘synthetic long peptides’ that promote adaptive immunity by authentic proteolytic processing and efficient antigen presentation.

Small circRNA elicited pro-inflammatory innate immunity and T cell responses. RNA-seq transcriptomic analysis verified that small circRNA activated a series of PRRs (for example, RIG-I), which provides the basis to further elucidate the innate immunomodulatory mechanisms for small circRNA vaccines. The ability and mechanisms of innate immunomodulation by small circRNA can be likely impacted by contexts, such as circRNA sequence and structure, nucleoside modifications, circRNA doses and pharmacokinetics, cell types and delivery systems. Compared with unmodified mRNA, which is typically highly immunostimulatory, unmodified small circRNA appears less immunostimulatory (comparably immunostimulatory relative to 5moU-modified mRNA) and more biostable, which is expected to permit long-lasting peptide expression by small circRNA. It is worth noting that circRNA and ionizable LNPs collectively mediate innate immunomodulation, which can promote antitumour and antiviral pro-inflammatory immunity but may also aggravate adverse side effects. Thus, it is desired to balance between the immunostimulation and the potential reactogenicity caused by circRNA LNPs for optimal clinical outcomes. In benchmarking studies, small circRNA vaccines outperformed several protein/peptide-encoding modified mRNA, unmodified mRNA and protein-encoding large circRNA to elicit potent and long-lasting antigen-specific T cell responses in mice. The corresponding PBMC and splenic antigen-specific T cell counts elicited by 3 × 10 μg small circRNA vaccines are estimated to be equivalent to the typical dose of adoptive TCR-T cells for mouse solid tumour immunotherapy. This provides an approach for in vivo production of therapeutically significant doses of antitumour TCR-T cells that can be boosted when needed, without any complex, costly and time-consuming ex vivo TCR-T cell engineering and manufacturing. Immune memory is pivotal for the long-term disease prophylactic and therapeutic response, as evidenced by ~1% chimeric antigen receptor (CAR) T cells in cured patients with cancer even years after CAR-T cell therapy⁴³. Small circRNA vaccines elicited great T cell memory with large fractions of antigen-specific T cells that persisted for at least 6 months post immunization, which outperformed benchmark mRNA vaccines. Moreover, small circRNA elicited profound T cell response in immunosenescent aged mice. The ability of small circRNA vaccine to elicit potent and long-lasting T cell response was attributed to the high biostability of circRNA that prolonged its in vivo half-life and the synchronized innate immunostimulation and efficient antigen production over a prolonged duration. Lastly, compared with a modified mRNA and large circRNA, small circRNA showed low PKR activation, high cell viability and superior safety in mice.

Small circRNA vaccines can be easily adjusted to encode various types of peptide antigen for broad applications. Small circRNA encoding MHC-I-restricted or MHC-II-restricted antigens elicited CD8⁺ or CD4⁺ T cells, both pivotal for the prophylaxis and therapy of cancer and infectious diseases. Moreover, small circRNA vaccines elicited potent T cell responses against various disease-associated antigens such as tumour-associated antigens, tumour neoantigens, viral antigens and oncoviral antigens. Furthermore, small circRNA can be designed to express multiple antigens, which is critical for optimal tumour therapeutic efficacy by overcoming tumour antigenic heterogeneity and minimizing tumour immune evasion. The applicability of crTMV IRES-based circRNA in various antigens suggests the high structural integrity of this IRES in circRNA. We envision that engineering structurally rigid and strong ribosome-binding small IRES holds the potential to develop small circRNA for efficient peptide production in versatile applications. In mice, circRNA vaccines, especially when combined with ICB, remodelled the tumour immune microenvironment as indicated by the reduced frequencies of tumour immunosuppressive cells, enhanced tumour infiltration of antitumour immune cells and the overall transcriptomic shift to an antitumour pro-inflammatory status. Low-dose mono-/multivalent small circRNA vaccines exhibited robust tumour immunotherapy and enhanced the ICB therapeutic efficacy, including 50% of complete regression of ICB-resistant Braf^V600E melanoma without recurrence in 2 months. Overall, small circRNA vaccines exhibited marked potential for tumour immunotherapy.

Compared with mRNA or large circRNA, small circRNA may hold advantages, including high stability, potent and long-lasting T cell responses, low PKR activation and good safety potential. Small circRNA vaccines may also find some limitations. For example, their application may be limited for large antigens that are often required to elicit humoral responses or for encoding many peptide antigens in the same circRNA vector to elicit multivalent T cell responses. Furthermore, owing to the limited length of RNA oligonucleotides synthesized by current automated RNA synthesizers, the synthesis of longer circRNA by ligating an increasing number of RNA oligonucleotides increases the complexity of circRNA manufacturing and may need to use alternative circRNA synthesis approaches (such as PIE systems⁴⁴). Lastly, our results could spark future studies to further understand and develop small circRNA vaccines, including IRES–ribosome interaction in circRNA, peptide translation mechanisms, the immunological mechanisms underlying the robust and long-last immunity, and expanded applications such as infectious disease prevention and therapy.

Methods

circRNA synthesis

All oligonucleotides (sequences in Supplementary Table 1) were purchased from Integrated DNA Technologies. Small circRNAs were synthesized using T4 RNA ligase 1 or 2 (New England Biolabs) to ligate linear 5′-phosphorylated RNA precursors (5 μM each) in the presence of 30-mer DNA splints (15 μM each). The products were treated with DNase I and exonuclease T (New England Biolabs) or RNase R (Abcam) to remove DNA and linear or lariat RNA. circRNA-RBD and circRNA-OVA (sequence in Supplementary Table 1) were synthesized in two steps as reported previously⁴⁴: (1) IVT using AmpliScribe T7 High Yield Transcription Kit (Biosearch Technologies, AS3107) and (2) group I intron-mediated autocatalytic circularization. In brief, 1 μg DNA template was used per 20 μl IVT reaction (37 °C, 2 h). Then, DNA templates were degraded with DNase I (37 °C, 15 min). The remaining RNA was column purified. Next, GTP (2 mM) was added to the resulting RNA to allow autocatalytic RNA cyclization (55 °C, 15 min). circRNA was column purified using an RNA Clean & Concentrator kit (Zymo Research) or HPLC to remove DNA or RNA fragments, nucleotides and enzymes. circRNA products were verified using gel electrophoresis, HPLC and Sanger sequencing of complementary DNA reverse transcribed from circRNA using primers flanking the circularization sites. Primers for circRNA-SIINFEKL reverse transcription: F, 5′-TTCGTTTGCTTTTTGTAGTATAATT-3′; R, 5′-AGTTTTTCAAAGTTGATTATACTCTCC-3′. Primers for circRNA-RBD reverse transcription: F, 5′-AAGCGGCTACATCCCAGAAG-3′; R, 5′-GGCGCACAAAGGTACCGTGA-3′.

In vitro circRNA stability

To study thermostability, small circRNA, liRNA and 5moU-mRNA-OVA (Trilink BioTechnologies, LLC) were stored in PBS at 4 °C, 25 °C and −20 °C for a series of times. Remaining RNAs were resolved by agarose gel electrophoresis and imaged using a ChemiDoc imaging system (Bio-Rad Laboratories). Intact RNA was quantified using ImageJ software (NIH). To study the in vitro biostability of large circRNA and small circRNA, RNA was incubated in freshly prepared diluted FBS (37 °C, 30 min) and was then briefly denatured in RNA loading buffer before gel electrophoresis. Intact RNA was quantified as above.

RNA LNP preparation

An ethanol phase containing all lipids and an aqueous phase of RNA were mixed to synthesize LNPs. The ethanol phase contained ionizable lipids, DSPC, DMG-PEG₂₀₀₀ and cholesterol at a molar ratio of 50:10:1.5:38.5. The aqueous phase contained circRNA or mRNA in 10 mM citrate buffer. The two phases were mixed at a flow rate of 1.8 ml min⁻¹ and 0.6 ml min⁻¹ (3:1), respectively, using Pump33DS syringe pumps. LNPs were dialysed in 1× PBS in a microdialysis cassette (20,000 molecular weight cut-off (MWCO), Thermo Fisher Scientific) at 4 °C for 12 h. LNP diameters and polydispersity index were measured on Zetasizer Nano (Malvern Instruments). RNA concentration and encapsulation efficiency in LNPs were measured using a modified Quant-iT RiboGreen (ThermoFisher) assay, and the circRNA copy numbers per LNP were estimated as before⁴⁵. For cryo-EM, LNPs were concentrated after dialysis to ~90 mg ml⁻¹ of lipid using Amicon ultracentrifugation filters before cryo-EM observation at the Molecular Electron Microscopy Core of University of Virginia.

circRNA LNP stability

Circular or linear Broccoli RNA aptamer loaded in LNPs was stored in PBS at 4 °C and −20 °C (supplemented with sucrose) for a series of durations. Then, recovered LNPs were transfected into DC2.4 cells for 1 h and were then resuspended and further incubated in PBS with 5% FBS and 200 μM DFHBI-1T (Sigma-Aldrich) for 45 min. Using untransfected cells as a control, the Broccoli fluorescence intensity (excitation, 488 nm; emission, 525 nm) of cells was analysed by flow cytometry.

Liposome encapsulation of circRNA

DOTAP, cholesterol and DSPE-PEG₂₀₀₀ (Avanti Polar Lipids) were mixed at molar ratios of 1:1:0.1 of DOTAP/cholesterol/DSPE-PEG₂₀₀₀. Lipid mixtures were dried under a nitrogen stream in glass test tubes. The resulting lipid films were placed in a desiccation system overnight. Dry films were hydrated in a solution of circRNA or mRNA in PBS with six cycles of vortexing for 30 s every 5 min. Samples were then freeze-thawed six times between liquid nitrogen and a 37 °C water bath and extruded using 21 passes in a mini-extruder (Avanti) through a 0.2 μm pore size polycarbonate filter (Whatman). Samples were placed in 10 K MWCO dialysis cartridges (Thermo Scientific) and dialysed against PBS for 6 h before use.

Cell culture

EG7.OVA cells were cultured in RPMI-1640 medium with 2 mM l-glutamine, 10% heat-inactivated FBS, 100 U ml⁻¹ penicillin, 100 μg ml⁻¹ streptomycin and 0.4 mg ml⁻¹ G418. DC2.4, TC-1, SM-1 and B3Z cells were cultured in RPMI-1640 medium (Gibco) supplemented with 10% FBS, 100 U ml⁻¹ penicillin, 100 μg ml⁻¹ streptomycin, 2 mM l-glutamine, 50 μM 2-mercaptoethanol, 1× non-essential amino acids and 10 mM HEPES. DC2.4 cells were cultured in RPMI-1640 medium supplemented with 10% FBS, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin. B16F10, MC38 and HEK293T cells were cultured in DMEM supplemented with 10% FBS, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin. All cells were cultured at 37 °C with 5% CO₂.

BMDC isolation and culture

Mouse BMDCs were isolated from C57BL/6 mice (6–8 weeks). All femurs and tibia were collected and cut at the epiphysis level, and bone marrows were flushed out with 10 ml RPMI-1640 medium (Gibco) 3 times. The resulting cell suspensions were filtered through cell strainers (70 μm cut-off, BD Falcon), centrifuged and then lysed by ACK buffer (Gibco). To culture BMDCs, RPMI-1640 medium supplemented with 10% FBS and 20 ng ml⁻¹ GM-CSF was added to resuspended cell pellets to obtain at a density of 2 × 10⁶ viable cells per 75 mm Petri dish. Three days later, an additional 10 ml cell culture medium was added. Six days later, non-adherent and loosely adherent cells were collected by gentle washing with PBS and then pooled for further studies.

Cell viability assay

The cytotoxicity of DC2.4 cells (2 × 10⁵ cells were seeded per well in 24-well plates) treated with small or large circRNAs was assessed using a ONE-Glo Tox kit (Promega, E7110). RNAs were transfected with Lipofectamine 3000 using serum-free medium 24 h after cell seeding. After 6 h, the medium was replaced by complete culture medium and the cells were incubated for another 18 h, before adding CellTiter-Fluor reagent. After incubation for 0.5 h, the fluorescence of cells was measured using a plate reader (Agilent BioTek) (excitation, 400 nm; emission, 505 nm).

Immunostaining of circRNA translation products in live DCs

Flag-encoding liRNA or circRNA (1 μg) was transfected into pre-seeded DC2.4 cells using Lipofectamine 3000 (Invitrogen). After 24 h, cells were washed and fixed using 4% paraformaldehyde in PBS at room temperature for 30 min and then incubated with 0.3% Triton X-100 in PBS. After blocking with 1% bovine serum albumin in PBS, cells were incubated with 2 μg ml⁻¹ of anti-Flag M2 monoclonal antibody (Sigma-Aldrich) in PBS containing 1% bovine serum albumin for 1 h. Cells were washed twice with PBS containing 0.3% Triton X-100 and then incubated with a mixture of Alexa Fluor 488-labelled anti-mouse IgG antibody (Invitrogen) and 4′,6-diamidino-2-phenylindole (DAPI; 250 ng ml⁻¹) for 30 min. The cells were mounted on a glass slide using PermaFluor aqueous mounting medium (Thermo Fisher Scientific) for observation by confocal laser scanning microscopy on a Zeiss LSM 780 confocal microscope.

Intracellular translation kinetics of peptide and protein from mRNA and circRNA

DC2.4 cells were seeded into 24-well plates at a density of 2 × 10⁵ cells per well and cultured at 37 °C for 24 h. Then cells were transfected with 5moU-mRNA-fFuc or circRNA-Flag (1 μg ml⁻¹) using Lipofectamine 3000. For 5moU-mRNA-fFuc, after 24 h, 48 h and 72 h, the cell culture supernatants were replaced with a ONE-Glo Luciferase Assay System according to the manufacturer’s instructions, and the bioluminescence was measured using a microplate reader. For circRNA-Flag, the cells were collected, permeated and intracellularly stained with PE-conjugated anti-Flag antibody (BioLegend, 637309), followed by flow cytometric analysis of cell fluorescence intensity.

PKR activation

PKR activation was studied in mouse DC2.4 cells and human HEK293 cells. DC2.4 cells seeded in 24-well plates were treated with LNPs of circRNA-SIINFEKL or 5moU-mRNA-OVA (100 nM) for 24 h at 37 °C. Then, cells were collected and lysed with cell lysis buffer (Cell Signaling Technology) containing 100 mM phenylmethylsulfonyl fluoride (Thermo Fisher Scientific). Cell lysates were centrifuged at 12,000 rpm for 15 min at 4 °C and the supernatants were eluted with 1× SDS loading buffer and resolved by 10% SDS–PAGE, followed by western blotting using anti-PKR and anti-pPKR antibodies (Thermo Fisher Scientific). HEK293T cells were used to study PKR activation by small circRNA and large circRNA, with PBS and poly(I:C) as controls. Following RNA transfection, cells were lysed for 4 h, and proteins were resolved by SDS–PAGE, followed by western blotting to detect PKR and p-PKR using the above antibodies. Relative PKR phosphorylation was calculated as the gel density ratio of pPKR/β-actin, which was then normalized to the pPKR/β-actin gel density ratio in PBS-treated cells.

Flow cytometry

Flow cytometry was used to analyse immunostained cultured cells, mouse blood cells, as well as tumour tissue and lymph node homogenate single cells. Antibodies used for immunostaining are listed in Supplementary Table 3. Cultured cells were analysed on a BD Beckman Coulter flow cytometer. Blood and tissue single cells were analysed on LSRFortessa X-50 (BD Biosciences). Flow cytometry results were analysed using FlowJo V10 software. Representative flow cytometric analysis gating trees are shown in Supplementary Figs. 23–29.

Antigen presentation on cultured DCs

DC2.4 cells or BMDCs (2 × 10⁵) seeded in 24-well plates were treated with circRNA-SIINFEKL or controls for 24 h. Cells were then collected and stained with APC-labelled anti-mouse H-2K^b/SIINFEKL complex antibody (BioLegend). Cells were then washed and analysed by flow cytometry.

B3Z cell activation

Upon recognition of the H-2K^b/SIINFEKL complex, B3Z cells (engineered SIINFEKL-specific CD8⁺ T cells) are activated and produce β-galactosidase. The activation level of the B3Z cells can be measured by the absorption of B3Z cell solution upon β-galactosidase treatment. Specifically, DCs were treated with SIINFEKL-specific vaccines for 16 h (RNA vaccine (1 μg) was transfected using Lipofectamine 3000). Cells were subsequently co-cultured with B3Z cells for another 24 h and were then lysed for 4 h at 37 °C with a lysis buffer (PBS with 100 mM 2-mercaptoethanol, 9 mM MgCl₂, 0.2% Triton X-100 and 0.15 mM chlorophenol red-β-d-galactopyranoside). The reaction was stopped using a stop buffer (1 M sodium carbonate). B3Z T cell activation was quantified by measuring the absorbance at 570 nm with 635 nm as a reference wavelength. The B3Z cell activation is shown as the normalized optical density (OD) relative to the untreated cell control.

ELISA

DCs were plated at densities of 5 × 10⁵ cells per well in 6-well plates. Cells were treated with the indicated formulations for 24 h. RNA (1 μg) was transfected into the cells using Lipofectamine 3000. Secreted cytokines (such as IFNβ, IL-6 and IL-12) in cell culture medium were quantified using ELISA kits (R&D Systems).

PRR activity studies using reporter cells

HEK-Lucia RIG-I cells, HEK-Blue hTLR3 cells or HEK-Blue hTLR7 cells were respectively seeded into a 96-well plate at a density of 3 × 10⁴ cells per well and cultured at 37 °C for 24 h. Then cells were transfected with RNAs (1 μg ml⁻¹) using Lipofectamine 3000. After 24 h, cell culture supernatants were collected and treated with either QUANTI-Blue Solution (InvivoGen, rep-qbs) or QUANTI-Luc 4 Lucia (InvivoGen, rep-qlc4lg1) according to the manufacturer’s instructions. The bioluminescence or OD at 630 nm was measured using a microplate reader.

In vitro cell uptake of circRNA

In vitro cell uptake of dye-labelled circRNA was examined using confocal laser scanning microscopy and measured by flow cytometry. Specifically, circRNA was labelled with Cy5 via hybridization with a Cy5-modified 30-mer cDNA. Cy5-circRNA loaded in LNPs or liposomes were incubated with DC2.4 cells for a series of durations and stained with LysoTracker Green DND-26 (Life Technologies) and 10 μg ml⁻¹ Hoechst 33342 (Life Technologies) for 0.5 h. Cells were then washed with DPBS three times before confocal microscopy observation on a Zeiss LSM 780 confocal microscope. Alternatively, DC2.4 cells seeded in 24-well plates were treated with Cy5-labelled circRNA formulations as above, followed by flow cytometric analysis as above.

DC co-stimulation

Co-stimulatory factors CD80 and CD86 were stained and measured on vaccine-treated DC2.4 cells by flow cytometry. In brief, DC2.4 cells were treated with circRNA vaccine or controls, dissociated using non-enzymatic dissociation buffer, washed with PBS and resuspended in PBS with dye-labelled anti-mouse CD80 and CD86 antibodies at 4 °C for 30 min. Cells were then washed with PBS before flow cytometric analysis as above.

RNA-seq

Total RNA was isolated using AllPrep DNA/RNA Mini Kit (QIAGEN) with in-column DNase 1 digestion to obtain DNA-free RNA. The integrity of extracted RNA was analysed using the Agilent RNA 6000 Pico Kit (Agilent). An aliquot of 1 μg total RNA (RIN > 8) was used to prepare mRNA-seq libraries following the instruction of Illumina stranded mRNA-seq preparation Guide (Illumina). In brief, mRNA was captured by oligo (dT) magnetic beads from total RNA, cation fragmented and reverse transcribed into cDNA. cDNA was end repaired, ligated with RNA index anchor and PCR amplified to produce indexed libraries with IDT for illumined UD indexes. Following quality control, the resulting barcoded libraries were pooled in equal molarities and were paired-end (2 × 150 bp) sequenced on the NextSeq 2000 instrument (Illumina). The quality of RNA-seq reads was assessed with FastQC v0.11.9. Reads from individual samples were aligned using STAR aligner version 2.7.6a to mouse reference genome GRCm39. Raw gene counts of mapped reads were aggregated using featureCounts. The differential gene expression analysis was performed with Bioconductor package DESeq2 v1.30.0 using the normalized and filtered counts per gene from the RNA-seq data. Differentially expressed genes with adjusted P < 0.05 were subject to Gene Ontology and KEGG pathway enrichment analyses using the GSEA. The visualization of upregulated genes and associated Gene Ontology functions when comparing the three biological conditions was achieved using the Triwise R package. The Triwise R package enables two-dimensional visualization of differential gene expression patterns when comparing three biological conditions. It filters and calculates the average expression in the three biological conditions for each gene and then transforms this gene expression matrix to barycentric coordinates, reducing the three-dimensional matrix by one and retaining only the expression changes between samples. Limma package is then used to determine differentially expressed genes. These barycentric coordinates can then be plotted in a two-dimensional dot plot where the direction of a gene (a single dot) indicates in which condition(s) the gene is upregulated, while the distance from the origin represents the strength of upregulation.

circRNA-seq and circRNA variant calling for circRNA authenticity analysis

circRNA-seq was conducted for circRNA-SIINFEKL that was extracted from transfected cells using AllPrep DNA/RNA/miRNA Universal Kit (QIAGEN), with untransfected circRNA-SIINFEKL as a control. circRNA libraries were constructed using 2 μg total RNA (RIN > 7), and ribosomal RNA was depleted using Ribo-Zero as well as an enzymatic depletion method (Illumina) and was purified with AMPure XP Beads (Beckman Coulter). Linear RNA was then digested using RNase R (Biosearch Technologies). The resulting circRNA was purified with AMPure XP Beads. circRNA was then cation fragmented at 95 °C for 2 min. Next, reverse transcription and library construction were performed using Illumina Stranded Total RNA Prep, followed by ligation with Ribo-Zero Plus Sample Preparation Guide (Illumina). Final libraries were size analysis by Agilent HS DNA chips (Agilent) and quantified by Qubit dsDNA HS Assay (Invitrogen). Following quality control, the resulting barcoded libraries were pooled in equal molarities and were paired-end (2 × 150 bp) sequenced on a NextSeq 2000 instrument (Illumina). The quality of raw FASTQ reads was assessed using FastQC v0.11.9. For read mapping, a reference fasta composed of two copies of circRNA-SIINFEKL sequence less one nucleotide was used. Reads were mapped either in a paired-end mode (PE) or in a single-end (SE) mode following read merging step. Here the reads were merged with either bbmerge function from BBTools package or merged reads (FLASh). The mapping step was performed with bowtie2 (bt2) with default settings, as well as with NextGenMap setting a constrain of minimum 90% identity (-i 0.9) over 75% of read length (-R 0.75) for a read to be considered as mapped. Finally, the variant calling step was done using the LoFreq tool. In brief, the mapping errors were corrected using ‘lofreq viterbi’, followed by indel quality insertion with ‘lofreq indelqual’, followed by base and indel alignment qualities insertions with ‘lofreq alnqual’. The final calling step was performed using ‘lofreq call’ with options set to ‘--no-default-filter --bonf 1 --sig 1 --call-indels’. To calculate the nucleotide authenticity rate of circRNA, reference fasta was divided into left and right flanks. Allele frequencies of left and right flank variants detected with LoFreq were summed and divided by the length of the flank. The resulting value was then used to calculate the nucleotide rates of circRNA.

Animal studies

All animal work was conducted in compliance to the Guide for the Care and Use of Animals under protocols approved by the Institutional Animal Care and Use Committee (IACUC) at Virginia Commonwealth University and University of Michigan. Vaccines were injected into mice subcutaneously either into mouse footpad for vaccine delivery study or into the mouse tail base for immune analysis or therapy studies. Both subcutaneous and intramuscular vaccine administrations expected to elicit comparable magnitude and quality of antigen-specific cellular or humoral immune responses⁴⁶.

In vivo tissue and cell distribution of circRNA in mice

To study in vivo tissue distribution of circRNA, circRNA was labelled with IR800 via a 30-mer cDNA. IR800-circRNA loaded in LNPs (0.5 nmol circRNA) were subcutaneously administered in the foot pad of albino C57BL/6 mice (6–8 weeks). A series of days later, mice were imaged for IR800 fluorescence using an IVIS Lumina system (PerkinElmer). In another cohort, 24 h after administration, major organs were isolated and imaged on an IVIS Lumina system. The images were processed using Living Image analysis software (PerkinElmer).

To study intranodal cell distribution of circRNA by single cell flow cytometric analysis, C57BL/6 mice were subcutaneously administered with nanoparticles of 0.5 nmol Cy5-labelled circRNA at tail base. Twenty-four hours later, draining inguinal lymph nodes were collected, and lymph node single cells were prepared by passing minced lymph nodes through 70 μm cell strainers. Cells were then washed with and resuspended in cell staining buffer for staining using the following antibodies (BioLegend): Brilliant Violet 421 anti-mouse CD45, Alexa Fluor 594 anti-mouse CD11c, FITC anti-mouse CD11b, APC/Cy7 anti-mouse CD8a, PerCP/Cy5.5 anti-mouse CD4, Brilliant Violet 605 anti-mouse CD103, PE/Cy7 anti-mouse CD205, APC/Cy7 anti-mouse F4/80, PE/Cy5 anti-mouse CD11b and APC anti-mouse NK1.1. Zombie Aqua Fixable Viability Kit (BioLegend) was used to stain dead cells. Cells were then washed and analysed by flow cytometry.

Serum cytokine and chemokine measurement by Luminex

C57BL/6 mice were immunized, and blood was collected under isoflurane anaesthesia 12 or 24 h later. Blood was centrifuged for 5 min at 13,000 rpm, and sera were collected and stored at −80 °C. Pre-selected panels of cytokines and chemokines in the above samples were measured by Luminex (University of Virginia Flow Cytometry Core).

Tetramer staining on T cells

Mouse CD8⁺ and CD4⁺ T cells were stained for antigen-specific tetramers using PE-conjugated tetramers (NIH Tetramer Core Facility). In brief, peripheral blood was collected from the vaccinated mice and blood cells were enriched by centrifugation. Red blood cells were lysed using ACK lysis buffer for 10 min at room temperature. Blood clots were removed using a filter. Cells were washed twice in PBS and then stained using a Zombie Aqua Fixable Viability Kit (BioLegend) for 20 min at room temperature. Staining was quenched and cells were washed with FCS buffer (PBS buffer with 0.1% FBS). Cells were then blocked with anti-CD16/CD32 (BioLegend) for 10 min, followed by adding staining cocktail (PerCP/Cy5.5 anti-mouse CD3, APC/Cy7 anti-mouse CD8a, Brilliant Violet 421 anti-mouse PD-1 and tetramer-PE) and staining at room temperature for 30 min. Cells were then washed and 100 μl Cytofix was added into each well to resuspend cells before fixation at 4 °C for 20 min. Cells were then washed with Perm/Wash buffer and resuspended for flow cytometric analysis.

Intracellular cytokine staining in T cells

Peripheral blood was collected from immunized mice. Red blood cells were removed using ACK lysis buffer, and the obtained lymphocytes were transferred into U-bottom 96-well plates in 200 μl T cell culture media (RPMI-1640, 10% FBS, 100 U ml⁻¹ penicillin/streptomycin, 50 μM β-mercaptoethanol, 1× MEM non-essential amino acid solution and 1 mM sodium pyruvate). Lymphocytes were pulsed with antigen peptides (40 μg ml⁻¹) for 4 h, followed by addition of GolgiPlug Protein Transport Inhibitor containing brefeldin A (Thermo Fisher Scientific). The sequences of antigen peptides (CSBio) are shown in Supplementary Table 4. Cells were then placed in a culture incubator for 6 h before incubation with anti-CD16/CD32 for 10 min at room temperature. Cells were stained with APC/Cy7 anti-mouse CD8a, PerCP/Cy5.5 anti-mouse CD4 and Zombie Aqua Fixable Viability Kit for 20 min at room temperature. Cells were washed and subsequently fixed using Cytofix (BD Biosciences), washed, and permeabilized in 200 μl Cytoperm solution (BD Biosciences). Cells were then washed using Perm/Wash buffer (BD Biosciences), and permeabilized cells were then stained using PE anti-mouse IFNγ (BioLegend) and FITC anti-mouse TNF (BioLegend). Stained cells were washed for flow cytometric analysis.

Immune memory

Peripheral blood was collected from mice to analyse memory T cells. Immune memory was analysed by flow cytometric analysis of peripheral lymph node homing receptor, CD62L, and adhesion molecule, CD44. In brief, red blood cells were lysed using ACK lysis buffer, and blood cells were then collected by centrifugation and washing with FCS buffer. Cells were then blocked with anti-CD16/CD32 in FCS buffer for 10 min, followed by adding CD8-APC-Cy7, CD44-PE-Cy5, CD62L-FITC and dead cell-staining DAPI. Cells were stained at room temperature for 30 min and were then washed. Cytofix (100 μl) was added into each well to resuspend cells and cells were then incubated at 4 °C for 20 min. Cells were then washed with Perm/Wash buffer and resuspended for flow cytometric analysis. Central memory CD8⁺ T cells were analysed as CD44^hiCD62L^hi CD8⁺ T cells; effector memory CD8⁺ T cells had variable to low levels of CD62L and high CD44 levels; and naive CD8⁺ T cells had high levels of CD62L and low to intermediate levels of CD44. To verify immune memory, immunized mice were challenged by s.c. inoculation of tumour cells (3 × 10⁵) on the right shoulder. Tumour size was monitored every 3 days thereafter. Tumour volume was calculated using the formula volume = (length × width²)/2 and analysed using GraphPad Prism 7.

Tumour immunotherapy

About 3 × 10⁵ EG7.OVA, MC38, TC-1, B16F10 and SM-1 Braf^V600E cells, respectively, were subcutaneously inoculated on the shoulder of female C57BL/6 mice (6–8 weeks; The Jackson Laboratory; n = 6–8). Tumour growth was monitored by caliper measurement. Mice were euthanized when the maximal tumour dimension reached 2 cm, or the tumour volume exceeded 2,000 mm³, or developed ulceration. Mice were treated 6 days after tumour inoculation when tumours reached approximately 50 mm³. Typical doses: 5 μg RNA vaccine, 5 μg CpG + 10 μg antigen peptides or proteins, and 200 μg anti-PD-1 or anti-CTLA-4 (Bio X Cell, NH). Vaccine doses used in SM-1 Braf^V600E melanoma therapy: 30 μg circRNA, 15 μg circRNA CpG and 30 μg antigenic peptides. Vaccines were subcutaneously injected at mouse tail base to allow lymphatic draining every 6 days three times, and anti-PD-1 or anti-CTLA-4 was injected intraperitoneally every 3 days five times. For lymphocyte depletion, anti-CD4, anti-CD8 and anti-NK1.1 (200 μg) were intraperitoneally injected every 3 days five times. Tumour size and mouse weight were monitored every 3 days. Tumour volumes were calculated and analysed as described above. Results were analysed using GraphPad Prism 7.

Statistical analysis

Data represent mean ± s.e.m., unless denoted otherwise. Statistical analysis was performed in GraphPad Prism Software version 5.0. P values were calculated by one-way ANOVA with Bonferroni post-test, unless denoted otherwise.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The data supporting the results in this study are available within the paper and its Supplementary Information. Source data are available via figshare at https://doi.org/10.6084/m9.figshare.25919746. The raw and analysed datasets generated during the study are available for research purposes from the corresponding authors on reasonable request. Source data are provided with this paper.

Change history

12 August 2025
A Correction to this paper has been published: https://doi.org/10.1038/s41551-025-01490-w

References

Rappuoli, R. Bridging the knowledge gaps in vaccine design. Nat. Biotechnol. 25, 1361–1366 (2007).
Article CAS PubMed Google Scholar
Pardi, N., Hogan, M. J., Porter, F. W. & Weissman, D. mRNA vaccines—a new era in vaccinology. Nat. Rev. Drug Discov. 17, 261–279 (2018).
Article CAS PubMed PubMed Central Google Scholar
Sahin, U. et al. Personalized RNA mutanome vaccines mobilize poly-specific therapeutic immunity against cancer. Nature 547, 222–226 (2017).
Article CAS PubMed Google Scholar
Pascolo, S. Vaccines against COVID-19: priority to mRNA-based formulations. Cells 10, 2716 (2021).
Article CAS PubMed PubMed Central Google Scholar
Teijaro, J. R. & Farber, D. L. COVID-19 vaccines: modes of immune activation and future challenges. Nat. Rev. Immunol. 21, 195–197 (2021).
Article CAS PubMed PubMed Central Google Scholar
Gholamalipour, Y., Karunanayake Mudiyanselage, A. & Martin, C. T. 3′ end additions by T7 RNA polymerase are RNA self-templated, distributive and diverse in character-RNA-Seq analyses. Nucleic Acids Res. 46, 9253–9263 (2018).
Article CAS PubMed PubMed Central Google Scholar
Frederickson, R. & Herzog, R. W. RNA-based vaccines and innate immune activation: not too hot and not too cold. Mol. Ther. 29, 1365–1366 (2021).
Article CAS PubMed PubMed Central Google Scholar
Schwartz, S. L. & Conn, G. L. RNA regulation of the antiviral protein 2′-5′-oligoadenylate synthetase. Wiley Interdiscip. Rev. RNA 10, e1534 (2019).
Article PubMed PubMed Central Google Scholar
Vogel, A. B. et al. Self-amplifying RNA vaccines give equivalent protection against influenza to mRNA vaccines but at much lower doses. Mol. Ther. 26, 446–455 (2018).
Article CAS PubMed Google Scholar
Deering, R. P., Kommareddy, S., Ulmer, J. B., Brito, L. A. & Geall, A. J. Nucleic acid vaccines: prospects for non-viral delivery of mRNA vaccines. Expert Opin. Drug Deliv. 11, 885–899 (2014).
Article CAS PubMed Google Scholar
Liu, X. et al. Circular RNA: an emerging frontier in RNA therapeutic targets, RNA therapeutics, and mRNA vaccines. J. Control. Release 348, 84–94 (2022).
Article CAS PubMed PubMed Central Google Scholar
Obi, P. & Chen, Y. G. The design and synthesis of circular RNAs. Methods 196, 85–103 (2021).
Article CAS PubMed PubMed Central Google Scholar
Litke, J. L. & Jaffrey, S. R. Highly efficient expression of circular RNA aptamers in cells using autocatalytic transcripts. Nat. Biotechnol. 37, 667–675 (2019).
Article CAS PubMed PubMed Central Google Scholar
Liu, B. et al. A split prime editor with untethered reverse transcriptase and circular RNA template. Nat. Biotechnol. 40, 1388–1393 (2022).
Article CAS PubMed Google Scholar
Yang, L., Wilusz, J. E. & Chen, L. L. Biogenesis and regulatory roles of circular RNAs. Annu. Rev. Cell Dev. Biol. 38, 263–289 (2022).
Article CAS PubMed PubMed Central Google Scholar
Chen, C.-y. & Sarnow, P. Initiation of protein synthesis by the eukaryotic translational apparatus on circular RNAs. Science 268, 415–417 (1995).
Article CAS PubMed Google Scholar
Qu, L. et al. Circular RNA vaccines against SARS-CoV-2 and emerging variants. Cell 185, 1728–1744 (2022).
Article CAS PubMed PubMed Central Google Scholar
Wesselhoeft, R. A., Kowalski, P. S. & Anderson, D. G. Engineering circular RNA for potent and stable translation in eukaryotic cells. Nat. Commun. 9, 2629 (2018).
Article PubMed PubMed Central Google Scholar
Chen, R. et al. Engineering circular RNA for enhanced protein production. Nat. Biotechnol. 41, 262–272 (2022).
Article PubMed PubMed Central Google Scholar
Katrekar, D. et al. Efficient in vitro and in vivo RNA editing via recruitment of endogenous ADARs using circular guide RNAs. Nat. Biotechnol. 40, 938–945 (2022).
Article CAS PubMed PubMed Central Google Scholar
Tatomer, D. C. & Wilusz, J. E. An unchartered journey for ribosomes: circumnavigating circular RNAs to produce proteins. Mol. Cell 66, 1–2 (2017).
Article CAS PubMed Google Scholar
Xiao, M. S. & Wilusz, J. E. An improved method for circular RNA purification using RNase R that efficiently removes linear RNAs containing G-quadruplexes or structured 3′ ends. Nucleic Acids Res. 47, 8755–8769 (2019).
Article CAS PubMed PubMed Central Google Scholar
Fan, X., Yang, Y., Chen, C. & Wang, Z. Pervasive translation of circular RNAs driven by short IRES-like elements. Nat. Commun. 13, 3751 (2022).
Article CAS PubMed PubMed Central Google Scholar
Chappell, S. A. & Mauro, V. P. The internal ribosome entry site (IRES) contained within the RNA-binding motif protein 3 (Rbm3) mRNA is composed of functionally distinct elements. J. Biol. Chem. 278, 33793–33800 (2003).
Article CAS PubMed Google Scholar
Li, P. W., Li, J., Timmerman, S. L., Krushel, L. A. & Martin, S. L. The dicistronic RNA from the mouse LINE-1 retrotransposon contains an internal ribosome entry site upstream of each ORF: implications for retrotransposition. Nucleic Acids Res. 34, 853–864 (2006).
Article CAS PubMed PubMed Central Google Scholar
Skulachev, M. V. et al. Internal initiation of translation directed by the 59-untranslated region of the tobamovirus subgenomic RNA I2. Virology 263, 139–154 (1999).
Article CAS PubMed Google Scholar
Filonov, G. S., Moon, J. D., Svensen, N. & Jaffrey, S. R. Broccoli: rapid selection of an RNA mimic of green fluorescent protein by fluorescence-based selection and directed evolution. J. Am. Chem. Soc. 136, 16299–16308 (2014).
Article CAS PubMed PubMed Central Google Scholar
Mangeat, B. et al. Broad antiretroviral defence by human APOBEC3G through lethal editing of nascent reverse transcripts. Nature 424, 99–103 (2003).
Article CAS PubMed Google Scholar
Roth, S. H., Levanon, E. Y. & Eisenberg, E. Genome-wide quantification of ADAR adenosine-to-inosine RNA editing activity. Nat. Methods 16, 1131–1138 (2019).
Article CAS PubMed Google Scholar
Li, Y. & Breaker, R. R. Kinetics of RNA degradation by specific base catalysis of transesterification involving the 2′-hydroxyl group. J. Am. Chem. Soc. 121, 5364–5372 (1999).
Article CAS Google Scholar
Thompson, J. E. et al. Limits to catalysis by ribonuclease A. Bioorg. Chem. 23, 471–481 (1995).
Article CAS PubMed PubMed Central Google Scholar
Hou, X., Zaks, T., Langer, R. & Dong, Y. Lipid nanoparticles for mRNA delivery. Nat. Rev. Mater. 6, 1078–1094 (2021).
Article CAS PubMed PubMed Central Google Scholar
Kranz, L. M. et al. Systemic RNA delivery to dendritic cells exploits antiviral defence for cancer immunotherapy. Nature 534, 396–401 (2016).
Article PubMed Google Scholar
Lamers, M. M., van den Hoogen, B. G. & Haagmans, B. L. ADAR1: “editor-in-chief” of cytoplasmic innate immunity. Front. Immunol. 10, 1763 (2019).
Article CAS PubMed PubMed Central Google Scholar
Arnaud, M. et al. Sensitive identification of neoantigens and cognate TCRs in human solid tumors. Nat. Biotechnol. 40, 656–660 (2022).
Article CAS PubMed Google Scholar
Yousefzadeh, M. J. et al. An aged immune system drives senescence and ageing of solid organs. Nature 594, 100–105 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wang, J. C. & Livingstone, A. M. Cutting edge: CD4⁺ T cell help can be essential for primary CD8⁺ T-cell responses in vivo. J. Immunol. 171, 6339–6343 (2003).
Article CAS PubMed Google Scholar
Yadav, M. et al. Predicting immunogenic tumour mutations by combining mass spectrometry and exome sequencing. Nature 515, 572–576 (2014).
Article CAS PubMed Google Scholar
Moynihan, K. D. et al. Eradication of large established tumors in mice by combination immunotherapy that engages innate and adaptive immune responses. Nat. Med. 22, 1402–1410 (2016).
Article CAS PubMed PubMed Central Google Scholar
van Stipdonk, M. J. et al. Design of agonistic altered peptides for the robust induction of CTL directed towards H-2Db in complex with the melanoma-associated epitope gp100. Cancer Res. 69, 7784–7792 (2009).
Article PubMed Google Scholar
Nagarsheth, N. B. et al. TCR-engineered T cells targeting E7 for patients with metastatic HPV-associated epithelial cancers. Nat. Med. 27, 419–425 (2021).
Article CAS PubMed PubMed Central Google Scholar
Mangkalaphiban, K. et al. Transcriptome-wide investigation of stop codon readthrough in Saccharomyces cerevisiae. PLoS Genet. 17, e1009538 (2021).
Article CAS PubMed PubMed Central Google Scholar
Fraietta, J. A. et al. Determinants of response and resistance to CD19 chimeric antigen receptor (CAR) T cell therapy of chronic lymphocytic leukemia. Nat. Med. 24, 563–571 (2018).
Article CAS PubMed PubMed Central Google Scholar
Wesselhoeft, R. A. et al. RNA circularization diminishes immunogenicity and can extend translation duration in vivo. Mol. Cell 74, 508–520 (2019).
Article CAS PubMed PubMed Central Google Scholar
Carrasco, M. J. et al. Ionization and structural properties of mRNA lipid nanoparticles influence expression in intramuscular and intravascular administration. Commun. Biol. 4, 956 (2021).
Article CAS PubMed PubMed Central Google Scholar
Ols, S. et al. Route of vaccine administration alters antigen trafficking but not innate or adaptive immunity. Cell Rep. 30, 3964–3971 (2020).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

G.Z. acknowledges funding support from NIH (R01CA266981, R01AI168684, R35GM143014 and R21NS114455), DoD CDMRP Breast Cancer Breakthrough Award Level II (BC210931/P1), American Cancer Society Research Scholar Grant (RSG-22-055-01-IBCD), University of Michigan Rogel Cancer Center Discovery Award, NIH-NCATS KL2 scholarship (KL2TR002648) via Virginia Commonwealth University (VCU) C. Kenneth and Dianne Wright Center for Clinical and Translational Research (UL1TR002649), VCU Massey Cancer Center Molecules to Medicine pilot grant and VCU Commercialization Fund. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Microscopy was performed at the VCU Microscopy Facility, supported in part by funding from NINDS Center Core Grant 5 P30 NS047463 and, in part, by funding from the NCI Cancer Center Support Grant P30 CA016059. Flow cytometry was performed at the VCU Massey Cancer Center Flow Cytometry Shared Resource, which is supported, in part, with funding from NCI Cancer Center Support Grant P30 CA016059. We thank the University of Michigan Flow Cytometry Core, the ULAM (Unit for Laboratory Animal Medicine) In Vivo Animal Core (IVAC), and the Advanced Genomics Core for service. We thank A. Ribas, J. Moon, C. Wu and NIH Tetramer Core for kindly providing cells and tetramer reagents.

Author information

These authors contributed equally: Yu Zhang, Xiang Liu.

Authors and Affiliations

Department of Pharmaceutical Sciences, College of Pharmacy, University of Michigan, Ann Arbor, MI, USA
Yu Zhang, Xiang Liu, Qiyan Wang, Shurong Zhou, Suling Yang, Shimiao Liao, Lei Mei & Guizhi Zhu
The Key Laboratory of Zhejiang Province for Aptamers and Theranostics, Hangzhou Institute of Medicine (HIM), Chinese Academy of Sciences, Hangzhou, China
Yu Zhang & Tingting Shen
College of Biological Science and Medical Engineering, Donghua University, Shanghai, China
Ting Su
Department of Biostatistics, School of Medicine; Bioinformatics Shared Resource, Massey Cancer Center, Virginia Commonwealth University, Richmond, VA, USA
Bei Zhang, Khoa Huynh, Katarzyna M. Tyc, Xufeng Qu & Jinze Liu
Genomics Core, Virginia Commonwealth University, Richmond, VA, USA
Linying Xie
Department of Medicinal Chemistry, School of Pharmacy, Virginia Commonwealth University, Richmond, VA, USA
Youzhong Guo
Department of Human and Molecular Genetics and Institute of Molecular Medicine, School of Medicine; Massey Cancer Center, Virginia Commonwealth University, Richmond, VA, USA
Chunqing Guo & Xiang-Yang Wang
Bioinnovations in Brain Cancer, Biointerfaces Institute; The Developmental Therapeutics Program, Rogel Cancer Center; Center for RNA Biomedicine; MI-AORTA program, Frankel Cardiovascular Center, University of Michigan, Ann Arbor, MI, USA
Guizhi Zhu

Authors

Yu Zhang
View author publications
Search author on:PubMed Google Scholar
Xiang Liu
View author publications
Search author on:PubMed Google Scholar
Tingting Shen
View author publications
Search author on:PubMed Google Scholar
Qiyan Wang
View author publications
Search author on:PubMed Google Scholar
Shurong Zhou
View author publications
Search author on:PubMed Google Scholar
Suling Yang
View author publications
Search author on:PubMed Google Scholar
Shimiao Liao
View author publications
Search author on:PubMed Google Scholar
Ting Su
View author publications
Search author on:PubMed Google Scholar
Lei Mei
View author publications
Search author on:PubMed Google Scholar
Bei Zhang
View author publications
Search author on:PubMed Google Scholar
Khoa Huynh
View author publications
Search author on:PubMed Google Scholar
Linying Xie
View author publications
Search author on:PubMed Google Scholar
Youzhong Guo
View author publications
Search author on:PubMed Google Scholar
Chunqing Guo
View author publications
Search author on:PubMed Google Scholar
Katarzyna M. Tyc
View author publications
Search author on:PubMed Google Scholar
Xufeng Qu
View author publications
Search author on:PubMed Google Scholar
Xiang-Yang Wang
View author publications
Search author on:PubMed Google Scholar
Jinze Liu
View author publications
Search author on:PubMed Google Scholar
Guizhi Zhu
View author publications
Search author on:PubMed Google Scholar

Contributions

G.Z. conceptualized the project. G.Z. and Y.Z. designed the project. Y.Z., X.L., Q.W., Tinging Shen, S.Z., S.L., S.Y., Ting Su, L.M., L.X., Y.G. and C.G. performed the experiments. Y.Z., X.L., X.-Y.W. and G.Z. analysed the data. Y.Z., B.Z., K.H., K.M.T., X.Q., J.L. and G.Z. analysed the RNA-seq and circRNA-seq data. Y.Z., X.L. and G.Z. drafted and revised the paper, with assistance from K.H., S.Z., K.M.T. and J.L. All authors reviewed the paper.

Corresponding author

Correspondence to Guizhi Zhu.

Ethics declarations

Competing interests

G.Z. and Y.Z. are listed as inventors in a related patent application (WO2022173730A1). G.Z. is a co-founder, central scientific officer and equity holder of AmpedRNA Biosciences, LLC. The other authors declare no competing interests.

Peer review

Peer review information

Nature Biomedical Engineering thanks Christopher Jewell, Bowen Li and Jinjun Shi for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Liposomes promote circRNA delivery and antigen presentation in DCs.

a, DLS data showing the size distribution of blank liposome and lipo-circRNA with different N/P ratios in PBS (solid) and 1% FBS (dotted). b, Zeta potential of blank liposome and lipo-circRNA. c-d, In vitro cell uptake of circRNA and antigen presentation in DCs mediated by lipo-circRNA. c, Flow cytometry results and d, MFI of the SIINFEKL presentation in SIINFEKL-circRNA-treated DC2.4 cells (24 h). SIINFEKL antigen presentation on DC2.4 cells that were treated with Lipofectamine 3000-transfected circRNA and lipo-circRNA with different N:P ratios, respectively. Ns: non-significant, *p < 0.05, ***p < 0.001, by one-way ANOVA with Bonferroni post-test. e-h, In vitro intracellular delivery of circRNA in DCs by liposome. e, Confocal microscopy images of DC2.4 cells treated with lipo-circRNA for 1 h, 3 h, or 6 h. Blue: nuclei stained with Hoechst33342. Green: endolysosome stained with LysoTracker Green DND-26. Red: Cy5-circRNA. Insets: close-up views of single cells. f, Flow cytometry results of DC2.4 cells incubated with Lipo-circRNA and free circRNA for different times. g, Flow cytometry results showing MFI of DCs incubated with free or Lipo-Cy5-circRNA. ****p < 0.0001, by t-test of the AUC. h, The signal ratio of Cy5-circRNA outside/inside (O/I) the endolysosome.

Source data

Extended Data Fig. 2 Liposomes promoted the delivery of circRNA to draining lymph nodes and to key intranodal APC subsets in mice.

a, Liposomes promoted the delivery of IR800-circRNA to draining popliteal lymph nodes (circled) in BALB/c mice (0.2 nmol, s.c. injected at foot pad). b, AUC of the radiance efficiency. **p < 0.01, by t-test of the AUC. c, Ex vivo fluorescence images of BALB/c mice after s.c. injected at tail base for 24 h. d, Radiance efficiency of major organs and inguinal lymph nodes after s.c. injection of IR800-circRNA at tail base for 24 h. He: heart; Li: liver; Sp: spleen; Lu: lung; Ki: kidney; and LN: inguinal lymph node. e, The frequency of circRNA⁺ cDCs and macrophages in inguinal lymph node tissues 24 h after s.c. injection at tail base.

Source data

Extended Data Fig. 3 Small circRNA-SIINFEKL elicited potent T-cell responses in young adult mice.

a, Study design. C57BL/6 mice (n = 5) were vaccinated on day 0 and day 14, and PBMC T cell responses were analyzed starting from day 21. b, Tetramer staining on day21 – day70 showed that circRNA-SIINFEKL elicited potent SIINFEKL-specific CD8⁺ T cell response in mice (n = 5) that outperformed current benchmarks 5moU-modified CleanCap mRNA-OVA and CpG-adjuvanted OVA. c, circRNA-SIINFEKL elicited SIINFEKL-specific T cell memory (day 70). Tem: effector memory T cell; Tcm: central memory T cell. d, circRNA-SIINFEKL upregulated PD-1 expression on SIINFEKL-specific CD8⁺ T cells on day21. e, PD-1 MFI on live PBMC CD8⁺ T cells. f, circRNA-SIINFEKL enhanced PD-1 expression levels (MFI) and frequencies on SIINFEKL⁺CD8⁺ T cells than that on total CD8⁺ T cells in peripheral blood on day 21, indicating immune exhaustion often resulting from chronic immunostimulation and providing an opportunity to combine circRNA vaccines with immune checkpoint blockade for optimal T cell responses (Two-tailed paired t test). g, circRNA-SIINFEKL enabled mice to resist 3×10⁵ EG7.OVA tumour challenge 70 days post-vaccination. Vaccine delivery by liposome, s.c. injected at tail base, 5 μg RNA, 2 nmole CpG, 20 μg OVA. *: relative to circRNA. h, Mouse body weights after tumour challenge. Data represent mean ± SD (b-e) and mean ± s.e.m. in other figure panels (n = 5). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, by one-way ANOVA with Bonferroni post-test unless denoted otherwise.

Source data

Extended Data Fig. 4 Peptide-encoding small circRNA vaccine elicit stronger and more durable T-cell responses than protein-encoding unmodified mRNA and large circRNA vaccines in young adult mice.

a, Design of T cell response study in mice. C57BL/6 mice (n = 4-5; 6-8 weeks) were immunized with circRNA-SIINFEKL, unmodified mRNA-OVA, and large circRNA code OVA protein at 3 μg and 10 μg doses, respectively. b, 120-day kinetics of the PBMC SIINFEKL-specific CD8⁺ T cell percentages in the above immunized mice, suggesting that circRNA-SIINFEKL elicited overall stronger and more durable T cell responses than OVA-encoding unmodified mRNA and large circRNA vaccines. Asterisks: statistical significance of the T cell fraction AUC relative to that for circRNA. c, circRNA-SIINFEKL elicited larger fractions of memory T cells (Tem + Tcm) than unmodified mRNAs (day 90), indicating great T cell memory elicited by small circRNA vaccine. Data were quantified from CD44 and CD62L staining of PBMC CD8⁺ T cells.

Source data

Extended Data Fig. 5 Monovalent small circRNA vs modified mRNA or large circRNA vaccines for robust tumour immunotherapy.

a, For immunotherapy studies in mouse models of subcutaneous EG7.OVA (b-d), and TC-1 (e), tumours were inoculated into the right flank of C57Bl/6 mice. Vaccine: 5 μg RNA or 5 μg CpG + 10 μg protein or peptide antigens, s.c. injected in liposome at mouse tail base; antibodies: 200 μg, i.p. b-d, EG7.OVA tumour growth (b) and Kaplan-Meier survival curves (c) of EG7.OVA tumour-bearing mice treated with circRNA-SIINFEKL, 5moU-mRNA-OVA, and CpG-adjuvanted OVA, respectively. αCD8, αCD4, and αNK1.1 were injected intraperitoneally (i.p.) for lymphocyte depletion. d, Body weights of EG7.OVA tumour-bearing mice treated with circRNA-SIINFEKL vs. controls. e, Individual (upper panel) and average (lower panel) EG7.OVA tumor growth curves in C57BL/6 mice treated with the indicated circRNA-SIINFEKL or large circRNA-OVA, as well as αPD-1 (i.p.) alone or combined with circRNA-SIINFEKL. RNA: 30 μg, s.c. injection at tail base. CR: complete regression rate. f, TC-1 tumour volumes after lymphocyte depletion using αCD8, αCD4 or αNK1.1. Data represent mean ± s.e.m. *p < 0.05; **p < 0.01; ***p < 0.001, by one-way ANOVA with a Bonferroni post-test (n = 6-8).

Source data

Supplementary information

Supplementary Information

Supplementary methods, figures, tables and references.

Reporting Summary

Source data

Source Data Figs. 1–8 and Source Data Extended Data Figs. 1–5

Source data and statistics.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Cite this article

Zhang, Y., Liu, X., Shen, T. et al. Small circular RNAs as vaccines for cancer immunotherapy. Nat. Biomed. Eng 9, 249–267 (2025). https://doi.org/10.1038/s41551-025-01344-5

Download citation

Received: 30 May 2023
Accepted: 07 January 2025
Published: 07 February 2025
Version of record: 07 February 2025
Issue date: February 2025
DOI: https://doi.org/10.1038/s41551-025-01344-5

This article is cited by

Circular RNA-based HPV16 therapeutic vaccine elicits potent and durable antitumor immunity
- Rong Zhou
- Chonghui Li
- Lu Gao
Journal of Experimental & Clinical Cancer Research (2026)
Circular RNAs in human biology: from splicing noise to master regulators
- Roya Kargar
- Emad Babakhanzadeh
Molecular Biology Reports (2026)
CircRNAs: functions and emerging roles in cancer and immunotherapy
- Yan Wang
- Yanru Cui
- Jian-Guo Zhou
BMC Medicine (2025)
Functions and applications of enzymes in nucleic acid nanotechnology
- Qinze Rong
- Zibin Chu
- Xin Su
Journal of Nanobiotechnology (2025)
Reprogramming neural-tumor crosstalk: emerging therapeutic dimensions and targeting strategies
- Qian-Qian Liu
- Zi-Kai Dong
- Wei-Lin Jin
Military Medical Research (2025)

Subjects

Abstract

Similar content being viewed by others

Main

Results

Optimization and characterization of highly stable small circRNA vaccines

LNPs efficiently delivered small circRNA to lymph nodes and APCs to elicit T cell responses

Small circRNA mediated immunomodulation with low PKR activation

Benchmarking small circRNA vaccines against modified mRNA, large circRNA and unmodified mRNA vaccines

Broad application of circRNA vaccines for various peptide immunogens

Low-dose circRNA vaccines reduced tumour immunosuppression for potent immunotherapy

Discussion

Methods

circRNA synthesis

In vitro circRNA stability

RNA LNP preparation

circRNA LNP stability

Liposome encapsulation of circRNA

Cell culture

BMDC isolation and culture

Cell viability assay

Immunostaining of circRNA translation products in live DCs

Intracellular translation kinetics of peptide and protein from mRNA and circRNA

PKR activation

Flow cytometry

Antigen presentation on cultured DCs

B3Z cell activation

ELISA

PRR activity studies using reporter cells

In vitro cell uptake of circRNA

DC co-stimulation

RNA-seq

circRNA-seq and circRNA variant calling for circRNA authenticity analysis

Animal studies

In vivo tissue and cell distribution of circRNA in mice

Serum cytokine and chemokine measurement by Luminex

Tetramer staining on T cells

Intracellular cytokine staining in T cells

Immune memory

Tumour immunotherapy

Statistical analysis

Reporting summary

Data availability

Change history

12 August 2025

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Extended data

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

This article is cited by

Search

Quick links