Abstract

   Synthetic circular RNA (circRNA) holds great promise for biomedical
   research and therapeutic applications, but impurities introduced during
   synthesis trigger innate immune responses and significantly compromise
   its efficacy. In this study, key immunogenic byproducts, including
   double‐stranded RNA, 5′ triphosphates from uncircularized RNA, and
   hydrolyzed RNA fragments, are identified as impairing circRNA
   functionality via RNA‐sensing pathways. To address this, a multi‐step
   purification process is developed that combines enzymatic treatments
   and cellulose‐based filtration to effectively remove these
   contaminants. This approach significantly reduces immune activation and
   increases manufacturing yields of circRNA by over 10‐fold. The purified
   circRNA demonstrates exceptional performance in induced pluripotent
   stem cells (iPSCs) reprogramming, achieving over 300% reprogramming
   efficiency with just three OSKMLN circRNA transfection treatments.
   Additionally, the purified circRNA achieves high levels and persistent
   expression of chimeric antigen receptor (CAR) in T cells with high
   cytotoxicity against tumor cells, making it a promising candidate for
   mRNA‐based CAR‐T cell therapy. These findings establish the
   purification strategy as a scalable and reliable platform for producing
   therapeutic‐grade RNA, with broad applications in mRNA‐based cell
   reprogramming for regenerative medicine and cancer immunotherapy.

   Keywords: cell reprogramming, circular RNA, iPSC, mRNA therapy,
   purification strategy
     __________________________________________________________________

   This study addresses the challenges of synthetic circular RNAimpurities
   that trigger immune responses and hinder efficacy. Key immunogenic
   byproducts are identified and develop a multi‐step purification process
   that produces ultra‐pure circRNA and enhances yields by over 10‐fold.
   The purified circRNA demonstrates remarkable reprogramming efficiency
   in iPSCs and high cytotoxicity in T cells, highlighting its potential
   for mRNA‐based reprogramming.

   graphic file with name ADVS-12-2416629-g007.jpg

1. Introduction

   Circular RNA (circRNA) is a type of RNA that forms a covalently closed
   loop, playing crucial roles in regulating gene expression and cellular
   processes.^[ [38]^1 , [39]^2 , [40]^3 , [41]^4 , [42]^5 ^] Recently, it
   has emerged as a promising therapeutic agent in fields such as cancer
   immunotherapy and vaccine development due to its unique structural
   properties.^[ [43]^6 , [44]^7 , [45]^8 , [46]^9 , [47]^10 , [48]^11 ^]
   Compared to linear mRNA, circRNA is relatively resistant to exonuclease
   degradation, resulting in enhanced stability. This stability allows for
   prolonged expression in target cells, making circRNA an ideal candidate
   for therapeutic applications.^[ [49]^12 ^]

   Previous studies have shown that mRNA‐based reprogramming offers
   several advantages, including low genotoxicity, high efficiency, and
   versatile delivery methods.^[ [50]^13 , [51]^14 ^] This process
   involves introducing synthetic mRNAs into target cells to translate
   proteins that induce pluripotency or direct differentiation. However,
   the rapid decay of synthetic mRNA often limits the effectiveness of
   this approach. In contrast, circRNA offers a compelling alternative as
   a result of its prolonged expression and stability, which can enhance
   the efficiency of cell reprogramming by maintaining protein levels over
   extended periods. Despite these benefits, the immunogenicity of circRNA
   remains controversial, as it may trigger immune responses that
   compromise cellular functions. Additionally, challenges in the
   synthesis and purification of circRNA can result in impurities that
   adversely affect its functionality and safety.^[ [52]^12 , [53]^15 ,
   [54]^16 , [55]^17 ^] Quality control methods, particularly ion‐paired
   reversed‐phase high‐performance liquid chromatography (IP‐RP HPLC),
   have been established to identify and characterize impurities in
   synthetic circular RNA, which can significantly impact the biological
   activities and overall efficacy.^[ [56]^18 ^] Addressing these issues
   is essential for unlocking the full potential of circRNA in the
   application of cell reprogramming.

   The production of synthetic circRNA generates various immunogenic
   byproducts during in vitro transcription (IVT) and RNA
   circularization.^[ [57]^15 ^] These impurities can activate multiple
   RNA‐sensing pathways, including Toll‐like receptors (TLRs) and
   RNA‐dependent protein kinase R (PKR), leading to immune responses that
   compromise RNA stability and expression.^[ [58]^19 , [59]^20 , [60]^21
   , [61]^22 , [62]^23 , [63]^24 , [64]^25 , [65]^26 , [66]^27 ^] Current
   purification methods, such as RNase R digestion and size‐exclusion
   liquid chromatography (SEC), are ineffective at removing structured or
   linear byproducts, resulting in immune activation and reduced protein
   expression.^[ [67]^15 , [68]^28 ^] While denaturing urea polyacrylamide
   gel electrophoresis can separate circRNA, it is unsuitable for
   large‐scale preparation. Therefore, optimized purification strategies
   are essential for producing high‐quality circRNA for therapeutic
   applications, making the development of scalable and efficient methods
   critical to realizing its full therapeutic potential.

   In this work, we systematically analyzed the immunogenicity of RNA
   byproducts generated during circRNA synthesis and identified four key
   contaminants: RNA precursors, nicked RNA, double‐stranded RNA (dsRNA),
   and RNA with 5′ triphosphate groups. Based on our findings, we
   developed a stepwise purification strategy that combines enzymatic
   digestion with cellulose filtration. This method effectively removes
   immune‐activating impurities, substantially reduces immune responses,
   and enhances circRNA expression, achieving superior manufacturing
   yields compared to conventional approaches. Most importantly, the
   purified circRNA demonstrated exceptional functionality and minimal
   immunogenicity in advanced therapeutic applications. In induced
   pluripotent stem cells (iPSCs) reprogramming and T cell engineering,
   our purified circRNA exhibited outstanding performance, highlighting
   its potential as a robust platform for regenerative medicine and cancer
   immunotherapy. These results establish our purification method as a
   promising approach for producing therapeutic grade circRNA suitable for
   medical applications.

2. Results

2.1. Byproducts in Synthetic circRNA Significantly Trigger Innate Immune
Responses

   In the synthesis of circRNA, transcription from a DNA template is
   followed by the circularization of linear precursors through either
   direct ligation with T4 DNA/RNA ligase or self‐splicing via group I/II
   autocatalytic introns.^[ [69]^6 , [70]^29 ^] The immunogenicity of
   circRNA can stem from the formation of dsRNA byproducts, uncircularized
   RNA (precursor), or nicked RNA molecules, all of which are potent
   triggers of cellular immune mechanisms (Figure [71]S1, Supporting
   Information).

   To identify the immunogenic RNA byproducts in synthesis circRNA, we
   employed a circRNA encoding enhanced green fluorescent protein
   (circ‐eGFP) according to ^(1) via group I intron‐based method. (Figure
   [72]1a). Capillary electrophoresis and IP‐RP HPLC analysis of the
   unpurified circRNA preparation revealed four distinct peaks: the low
   molecular weight peaks for excised introns of the circularization
   process, a main circRNA peak that contained the desired circRNA, a
   precursor peak, and a high molecular weight (HMW) peak (Figures [73]S2a
   and [74]S2c, Supporting Information). To explore the immunogenicity of
   different byproducts, we employed SEC to fractionate the unpurified
   circ‐eGFP into distinct components, including HMW, main, and intron
   fractions (Figure [75]1b). As shown in Figure [76]1c, the intron
   fraction contained introns and a small proportion of the main peak,
   whereas the main fraction contained circRNA, nicked RNA, and linear
   precursors. The HMW fraction primarily consisted of HMW byproducts such
   as dsRNA and other RNA aggregates.

Figure 1.

   Figure 1
   [77]Open in a new tab

   Innate immune activation by byproducts of IVT‐synthesized circular RNA.
   a) Schematic representation of RNA circularization with group I
   autocatalytic introns (Created with BioRender.com). b) Size exclusion
   chromatography separation of unpurified circRNA. c) Capillary
   electrophoresis of different fractions, as described in Figure [78]1b.
   d) Gene expression profiles of PMA‐differentiated THP‐1 cells
   transfected with different fractions (Data presented n = 3). e)
   Multiple KEGG enrichment analyses of PMA differentiated THP‐1
   transfected with different fractions, with untreated cells serving as
   the control group. f) Multiple volcano plots illustrating
   transcriptomic differences between THP‐1 cells transfected with
   different fractions. Untreated cells served as the control group. Each
   cluster represents the comparison of a specific fraction‐transfected
   condition against the control. g) Genes associated with innate immune
   response were identified as differentially expressed candidate mRNAs
   upon unpurified, intron, early and HMW transfection relative to
   untransfected cells (n = 3). B2 m mRNA was used as a housekeeping
   control. h) Induction of RIG‐I, TLR3/7/8, MDA5, PKR, and OAS1
   transcripts 6 h post‐transfection of PMA‐differentiated THP‐1 with
   unpurified, intron, main, and HMW fractions relative to untransfected
   cells (n = 3). B2 m mRNA was used as a housekeeping control. i–k)
   Induction of IFNβ, TNFα, and IL6 transcripts 6 h post‐transfection of
   PMA‐differentiated THP‐1 with unpurified, intron, main, and HMW
   relative to untransfected cells, B2 m mRNA was used as a housekeeping
   control. Data presented as means ± standard deviations (SDs), n = 3.

   We then transfected unpurified circRNA, main fraction, HMW, or intron
   fraction, into THP‐1 cells to examine their immunogenicity in cells
   (Methods). Transcriptome analysis revealed that cells treated with
   unpurified circRNA, as well as those treated with the main, intron, and
   HMW fractions, all exhibited significant changes in gene expression
   patterns (Figure [79]1d–f).

   Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment
   analysis further identified significant differences in immune response
   pathways between blank and various RNA fractions, with enriched
   pathways including TNF, TLR, NF‐κB signaling, and cytokine–cytokine
   receptor interactions. The unpurified cricRNA fraction showed the
   highest enrichment factors and significant p‐values, indicating strong
   pathway associations. The main fraction exhibited similar enrichment,
   likely due to linear precursors and nicked RNA. Despite lower
   transfection amounts, intron and HMW fractions also showed notable
   enrichment. We also identified genes upregulated by unpurified circRNA
   (Figure [80]1g), revealing significant activation of RNA sensors
   (TLR3/7/8, PKR, OAS, MDA5, RIG‐I) in unpurified, main, and HMW
   fractions, leading to cytokine and interferon production. The potent
   induction of IFNβ by the HMW fraction can likely be attributed to the
   high levels of dsRNA.^[ [81]^30 , [82]^31 , [83]^32 ^] RT‐qPCR results
   (Figure [84]1h–k) further confirmed these findings, showing that
   unpurified circ‐eGFP and HMW fractions had the highest induction of RNA
   sensors and cytokines, likely due to immunogenic RNA species such as
   dsRNA.

   These results reveal the heterogeneous nature of unpurified circRNA,
   suggesting that trace synthesis‐derived impurities can activate innate
   immune pathways and impair circRNA functionality, underscoring the need
   for effective purification strategies to ensure therapeutic safety and
   efficacy.

2.2. Removal of Immunogenic Contaminants to Enhance circRNA Performance

   Our findings indicate that trace high molecular weight (HMW)
   impurities, including dsRNA, can negatively impact circRNA
   functionality. To address this, we employed microcrystalline cellulose
   (MCC) chromatography to evaluate its efficacy in removing HMW
   impurities that adversely affect circRNA performance and protein
   expression (Figure [85]2a).^[ [86]^32 ^] Using an isopropanol
   (IPA)‐HEPES buffer system, we collected both the flow‐through and
   subsequent elution peaks (Figure [87]2b). Capillary electrophoresis
   revealed distinct RNA distributions: introns were primarily in the
   flow‐through and early elution peaks, while HMW species were mainly in
   the later elution peaks (Figure [88]S3a–c and Table [89]S1, Supporting
   Information). A lateral flow strip assay (LFSA) confirmed the absence
   of dsRNA in the early elution peaks, indicating successful removal
   (Figure [90]2c).^[ [91]^33 ^] Protein expression tests in HeLa cells
   showed that the initial elution peaks significantly enhanced protein
   expression and transfection efficiency, while later fractions exhibited
   diminished effects (Figure [92]2d; Figure [93]S3d, Supporting
   Information). Notably, the flow‐through fraction with the highest
   intron content also increased protein expression, underscoring the
   negative impact of HMW impurities, which were inversely correlated with
   protein expression outcomes (R^2 = 0.84, Figure [94]S3e, Supporting
   Information).

Figure 2.

   Figure 2
   [95]Open in a new tab

   Removal of immunogenic contaminants enhances circRNA translation
   efficiency and reduces immune response. a) Purification of IVT circRNA
   using MCC chromatography to eliminate dsRNA. (Created with
   BioRender.com). b) MCC chromatogram of unpurified circ‐eGFP utilizing
   20% (red) and 25% (blue) isopropanol (IPA)‐HEPES equilibrium
   conditions. c) LFSA assessment of circRNA fractions as outlined in
   Figure [96]2b. d) Representative cellular fluorescence image 24 h
   post‐transfection of HeLa cells with different fractions (scale bar:
   100 µm). e) Enzymatic treatment for immunogenic contaminant removal.
   (Created with BioRender.com). f) SEC separation of RR+ circ‐eGFP. g)
   Induction of RIGI, TLR3, TLR7, and MDA5 transcripts 6 h after
   transfection of A549 cells with the 40 ng of the intron, HMW, or 200 ng
   of a main fraction of SEC separated RR+circ‐eGFP (n = 3). B2 m mRNA was
   used as a housekeeping control. h) Induction of RIG‐I, TLR3, TLR7, and
   MDA5 transcripts 6 h after transfection of A549 cells with the 40 or
   200 ng of intron and HMW, before and after phosphatase treatment
   (intron‐P, HMW‐P), or 200 ng of main fraction after phosphatase
   treatment (main‐P) (n = 3). B2 m mRNA was used as a housekeeping
   control.

   To further enhance circRNA purity, we investigated enzymatic treatments
   to address immunogenic contaminants (Figure [97]2e). We subjected
   unpurified circ‐eGFP to RR treatment, which selectively degrades linear
   RNA while preserving circRNA. RR digestion achieved ≈89.8% circ‐eGFP
   purity (Figure [98]S2b–d, Supporting Information). However, the
   resulting protein expression levels were lower than those of samples
   treated solely for dsRNA removal (Figure [99]S4, Supporting
   Information). Testing RR‐treated circ‐eGFP in A549 cells revealed that
   dsRNA‐containing fractions elicited strong immune responses, while the
   RR‐treated main fraction, largely free of linear contaminants, showed
   reduced immunogenicity (Figure [100]2g,h).

   To address the immunogenicity of intron and HMW fractions, we treated
   them with phosphatase to remove 5′ triphosphate groups, known triggers
   for immune sensors like RIG‐I. Transfection into A549 cells showed that
   while phosphatase treatment reduced the immunogenicity of
   intron‐derived impurities, it had little effect on the immune response
   from HMW fractions, primarily containing dsRNA (Figure [101]S5,
   Supporting Information). The main fraction also exhibited low
   immunogenicity after phosphatase digestion, attributed to the removal
   of the remaining 5′ triphosphates.

2.3. Combined Strategy Removes Immunogenic Byproducts and Yields High‐Purity
circRNA

   We have identified several immunogenic byproducts in IVT‐synthesized
   circRNA, including dsRNA in HMW fractions, 5′ triphosphate introns,
   precursors, and nicked RNA. Effective removal of these impurities is
   critical to producing high‐purity, low‐immunogenicity circRNA, which is
   essential for ensuring optimal performance and safety in advanced
   therapeutic applications. To confirm that the contaminants are the
   major source of immunogenicity, we refined our purification protocol by
   integrating MCC with enzymatic treatments (Figure [102]3a). This
   approach aimed to achieve even higher purity levels of circRNA while
   mitigating immunogenic responses.

Figure 3.

   Figure 3
   [103]Open in a new tab

   An optimized purification strategy yields high‐purity circRNA with
   enhanced efficiency. a) Illustration of our optimized purification
   strategy (Created with BioRender.com). b) WMC separation of 200 µg of
   RR+circ‐eGFP using FPLC. c) LFSA test of RR+, RR+SEC‐L‐P, RR+WMC‐P‐1
   and RR+WMC‐P‐2. circ‐eGFP. d) Recovery rates following RR+SEC and
   RR+WMC (mean ± SD, n = 3). e,f) Protein expression levels of PBMCs 24 h
   post‐transfection with purified fractions (scale bar: 100 µm; One‐way
   ANOVA, ^* p < 0.05).

   We purified RR+circ‐eGFP using MCC, which was subsequently treated with
   phosphatase (RR+CC‐P). Capillary electrophoresis and LFSA results
   confirmed the effective removal of dsRNA from RR+CC‐P circ‐eGFP, with
   only a minor fraction of introns (0.8%) and HMW components (2.1%)
   remaining (Figures [104]S6a, S7a, and [105]S8, Supporting Information).
   For comparison, a stepwise process involving RR digestion, SEC, and
   phosphatase treatment was used as a control. Both the early and late
   peaks from SEC (RR+SEC‐E‐P, RR+SEC‐L‐P) were collected (Figure [106]S9,
   Supporting Information). Capillary electrophoresis and HPLC results
   confirmed the high purity of both RR+SEC‐E‐P and RR+SEC‐L‐P, although
   RR+SEC‐E‐P showed a slight increase in linear RNA. The RR+SEC‐L‐P
   fraction, obtained from the later SEC peak, was used as an ultra‐pure
   circRNA positive control (Figure [107]S8, Supporting Information;
   [108]Experimental Section ^[ [109]^15 ^]).

   We then evaluated purification efficacy by testing circRNA obtained
   through these different strategies. Protein expression in HeLa levels
   of RR+CC‐P circ‐eGFP were comparable to those of RR+SEC‐L‐P (ultra‐pure
   circRNA), demonstrating its effectiveness (Figure [110]S7b–d,
   Supporting Information). In contrast, RR+SEC‐E‐P performed
   suboptimally, likely due to its enrichment in linear RNA contaminants.

   Given these findings, we sought to further enhance the removal of
   immunogenic dsRNA while improving purification efficiency and
   scalability, a critical goal for large‐scale production. To achieve
   this, we introduced wood‐derived macroporous cellulose (WMC), a natural
   porous material, for the removal of dsRNA. The porous nature of WMC
   allows for efficient removal of dsRNA from circRNA of varying lengths
   while maintaining a high recovery rate of the target RNA on both
   laboratory‐ and industry‐scale.^[ [111]^34 ^]

   We evaluated the recovery efficiency of WMC chromatography using
   varying amounts of circ‐eGFP RNA (50, 200 µg, and 1 mg,
   Figure [112]3a,b). The 50 µg sample was purified using a
   micro‐separation column, while the 200 µg and 1 mg samples were
   processed using FPLC. Across all tested conditions, WMC completely
   removed dsRNA and achieved significantly higher recovery rates (>70%)
   compared to the conventional SEC‐based approach (Figure [113]3c,d),
   with particularly notable improvements at smaller input amounts (50 and
   200 µg).

   Following WMC or SEC purification, circ‐eGFP samples were treated with
   phosphatase, yielding final products designated as ultra‐pu6re:
   RR+WMC‐P‐1 (FPLC) and RR+WMC‐P‐2 (micro‐separation column),
   respectively. Purity assessments using capillary electrophoresis and
   HPLC confirmed that both WMC‐based methods produced high‐purity circRNA
   with negligible impurities (Figures [114]S7 and [115]S8, Supporting
   Information). Protein expression levels from WMC‐purified samples were
   comparable to those obtained using conventional protocols in HeLa cells
   and PBMCs (Figure [116]3e,f; Figure [117]S10, Supporting Information).
   These results indicate that WMC is an effective method for purifying
   diverse circRNA species, offering clear advantages in recovery rate and
   scalability compared to SEC‐based methods.

2.4. Purified circRNA Displays Minimal Immunogenicity

   Our studies demonstrate that circRNA purified using the RR+CC‐P method
   achieves protein expression levels comparable to those obtained through
   HPLC methods (e.g., RR+SEC‐L‐P). To explore the immunogenicity of our
   purified circRNA, we transfected THP‐1 cells with circ‐eGFP and a
   5moU‐modified linear eGFP. Transcriptional analysis revealed that
   RR+CC‐P circ‐eGFP exhibited significantly reduced immunogenicity, with
   levels comparable to those of RR+SEC‐L‐P (Figure [118]4a–d; Figure
   [119]S11, Supporting Information). In contrast, RR+SEC‐E‐P showed
   relatively higher immunogenicity, likely due to linear contaminants,
   which cannot be completely separated by SEC.

Figure 4.

   Figure 4
   [120]Open in a new tab

   Purified circRNA exhibits minimal immunogenicity. a) Expression levels
   of inflammatory cytokines (IFNβ, TNFα, and IL6) in PMA‐differentiated
   THP‐1 cells at 6‐ and 48‐h post‐transfection with circ‐eGFP (200 ng or
   1 µg) purified by different methods. Data normalized to B2 m and
   presented as mean ± SDs (n = 3). b) Expression levels
   of RIGI, IFNβ, IL6, and TLR3 in human primary fibroblasts at 6 and 48 h
   post‐transfection with circ‐eGFP purified by different methods. Data
   are normalized to B2m and presented as mean ± SD (n = 3). c) Expression
   levels of RIGI, IFNβ, IL6, and TNFα in PBMCs at 6 and 48 h
   post‐transfection with circ‐eGFP purified by different methods. Data
   are normalized to B2m and presented as mean ± SD (n = 3). d) Gene
   expression profiles of PMA‐differentiated THP‐1 cells transfected with
   indicated circ‐eGFP (n = 3). e) The volcano plot shows the difference
   in transcriptome between THP‐1 cells transfected with RR+SEC‐L‐P and
   RR+CC‐P circ‐eGFP.

   RNA sequencing experiments comparing unpurified, ultra‐pure, and
   RR+CC‐P circRNA further confirmed the low immunogenicity of our
   purified circRNA (Figure [121]4d,e). Both the RR+CC‐P and ultra‐pure
   (RR+SEC‐L‐P) circRNA groups exhibited minimal differential gene
   expression compared to untreated controls, while unpurified circRNA
   induced significant transcriptomic changes, particularly in innate
   immune response pathways.

   Despite achieving high purity and low immunogenicity with RR+CC‐P
   circRNA, trace amounts of non‐dsRNA HMW impurities and introns
   persisted. To further evaluate the influence of the remaining
   contaminants, we conducted a differential expression analysis. The
   results showed that cells treated with HMW impurities before MCC
   resulted in 2762 upregulated and 1619 downregulated genes compared to
   treatment after MCC (Figure [122]5a). Differentially expressed genes
   related to immune responses were identified (Figure [123]5b), with KEGG
   and GO analyses showing significant associations with TNF signaling,
   cytokine interactions, and TLR signaling (Figure [124]5c; Figure
   [125]S12, Supporting Information). Thus, while dsRNA removal reduces
   immune responses, residual impurities, possibly the dimer form of
   circ‐eGFP, may still play a role.

Figure 5.

   Figure 5
   [126]Open in a new tab

   RNA Sequencing reveals residual HMW species and introns do not induce
   significant immunogenicity. a) The volcano plot shows the difference in
   transcriptome between THP‐1 cells transfected with HMW relative to the
   HMW after MCC chromatography (HMW‐CC) treated cells. b) Genes
   associated with innate immune response were identified as
   differentially expressed candidate mRNAs upon HMW transfection relative
   to HMW‐CC treated cells (n = 3). B2 m mRNA was used as a housekeeping
   control. c) KEGG enrichment analysis of PMA differentiated THP‐1
   transfected with HMW‐CC, and HMW. d) The volcano plot shows the
   difference in transcriptome between THP‐1 cells transfected with
   intron‐P relative to the intron treated cells. e) Genes associated with
   innate immune response were identified as differentially expressed
   candidate mRNAs upon intron transfection relative to intron‐P treated
   cells (n = 3). B2 M mRNA was used as a housekeeping control. f) KEGG
   enrichment analysis of PMA differentiated THP‐1 transfected with intron
   and intron‐P.

   For intron fractions, transcriptome analysis indicated 363 genes were
   upregulated and 266 downregulated before and after phosphatase
   treatments (Figure [127]5d,e). KEGG and GO analyses (Figure [128]5f;
   [129]S13a, Supporting Information) showed these genes were mainly
   involved in the innate immune response. RT‐qPCR confirmed decreased
   mRNA levels of RNA molecular sensors and cytokines post‐phosphatase
   treatment (Figure [130]S13b,c, Supporting Information). Overall, these
   findings highlight the effectiveness of our purification strategy in
   reducing immunogenic responses.

2.5. Purified circRNA Enabling High‐Efficiency iPSCs Reprogramming

   Our findings demonstrate that even trace amounts of impurities can
   significantly impact protein expression and cellular function. Thus,
   achieving high‐purity circRNA is essential for optimal outcomes and
   preventing dysfunction related to circRNA impurities. To evaluate the
   efficacy of our highly purified circRNA, we conducted cellular
   reprogramming experiments using circRNA derived from the transcription
   factors OSKMLN (Oct4, Sox2, Klf4, cMyc, Nanog, and Lin28A).

   The activation of inflammatory pathways is one of the key factors in
   causing reprogramming failure.^[ [131]^35 , [132]^36 ^] Previous
   studies have shown that mRNA‐based approaches achieve high‐efficiency
   reprogramming as a result of low immunogenicity and high transfection
   efficiency.^[ [133]^13 ^] Our experiments revealed that transfection of
   RR+WMC‐P circRNA at varying frequencies from day 1 to day 7 resulted in
   the efficient generation of iPSC colonies with as few as two
   transfections (Figure [134]6a,b). Notably, the group receiving three
   transfections achieved over 3000 iPSC colonies, corresponding to a
   reprogramming efficiency exceeding 300%, a level typically only
   attainable with modified mRNA.^[ [135]^13 ^] We also observed that
   iPSC‐like colonies emerged earlier in groups transfected with
   ultra‐pure circRNA (day 12, Figure [136]S14, Supporting Information).

Figure 6.

   Figure 6
   [137]Open in a new tab

   Functional validation of purified circRNA in cell reprogramming. a) A
   schematic diagram illustrating the induction of iPSCs using circRNA
   (Created with BioRender.com). b) Representative images of
   TRA‐1‐60‐stained reprogramming wells, showing the yield of
   TRA‐1‐60‐positive colonies on day 14 under the RR+WMC‐P circRNA‐based
   reprogramming regimen with varying transfection frequencies, initiated
   with a plating density of 1000 cells. c) Immunofluorescence analysis of
   the three germ layers differentiated from the indicated iPSC lines
   showing histology and marker expression specific to ectoderm
   (βIII‐tubulin (TUJ1)), mesoderm (vimentin), and endoderm
   (α‐Fetoprotein). d) Karyotyping of indicated iPSC lines. e) RNA
   sequencing of fibroblasts, ^(1) iPSCs induced by the RR+WMC‐P circRNA,
   ^(2) and H9 iPSC positive controls, ^(3) with data shown as the mean of
   four biological replicates.

   To investigate transcriptional heterogeneity within the iPSC
   population, we performed RNA‐Seq analysis on four randomly selected
   lines of the circRNA‐induced iPSC colonies. The resulting
   transcriptomic profiles were compared with two control groups: the
   parental fibroblast population and the well‐characterized human
   embryonic stem cell line H9. As illustrated in Figure [138]6e, the
   iPSCs generated via circRNA reprogramming exhibit a global gene
   expression pattern consistent with that observed in human embryonic
   stem cells, particularly with respect to key pluripotency markers.^[
   [139]^13 , [140]^37 , [141]^38 ^] Moreover, there was a downregulation
   of fibroblast identity genes, indicating successful repression of the
   somatic transcriptional program and acquisition of a pluripotent state.

   In contrast, the suboptimal circRNA (RR+SEC‐E‐P) failed to induce the
   fibroblast‐to‐iPSC transition under identical conditions (1000 cells
   seeded, 2 or 3 transfections), resulting in no iPSC colonies. While
   increasing the number of transfections could yield some iPSCs, the
   efficiency remained significantly lower (data not shown), likely due to
   the high immunogenicity of the suboptimal circRNA (Figure [142]4b).
   Elevated immunogenicity can lead to altered cell fate decisions,
   ultimately resulting in reprogramming failure.

   These results underscore the critical importance of utilizing
   high‐purity circRNA to create a conducive environment for successful
   reprogramming and to ensure the generation of high‐quality iPSCs.
   Furthermore, our purification strategy minimizes the risk of unwanted
   cellular stress responses, enhancing the overall stability and
   functionality of the reprogrammed cells. Ultimately, our findings
   highlight that circRNA purity is a key factor in optimizing
   reprogramming protocols for biomedical applications.

2.6. Purified circRNA Enabling Enhanced T cell Engineering

   We subsequently tested our purified circRNA in a challenging cellular
   application: T cell engineering. The use of high‐purity,
   low‐immunogenic gene vectors is crucial in this context, as it
   minimizes unwanted immune responses that can compromise T cell function
   and efficacy. We evaluated the efficacy of WMC‐purified anti‐human CD19
   circ‐CAR in modifying T cells (Figure [143]7a). Transcriptional
   analysis has revealed that WMC‐purified circ‐CAR elicited minimal
   innate immune responses (Figure [144]4c) while maintaining significant
   protein expression, comparable to ultra‐pure reference samples. Flow
   cytometric analysis further confirmed that T cells transfected with
   WMC‐purified circ‐CAR exhibited sustained CAR expression levels
   comparable to those of ultra‐pure circ‐CAR (RR+SEC‐L‐P), significantly
   surpassing the expression levels observed with RR+ and RR+SEC‐E‐P
   (Figure [145]7b,c).

Figure 7.

   Figure 7
   [146]Open in a new tab

   Functional validation of purified circRNA in T cell engineering. a)
   Schematic illustrating CAR‐T cell engineering via circRNA (Created with
   BioRender.com). b,c) CAR expression profiles in T cells transfected
   with circRNA purified by various methods, with data presented as the
   mean of three biological replicates. d) In vitro cytotoxicity of CAR‐T
   cell products generated from circRNA purified by various methods,
   following co‐culture with Raji‐luc cells, demonstrating a range of
   effects based on the effector‐to‐target (E:T) ratio, with data shown as
   the mean of three biological replicates.

   Following this, we conducted in vitro cytotoxicity assays using
   Raji‐luc as the target cells, demonstrating that CAR‐T cells derived
   from the purified group exhibited superior killing compared to those
   generated using suboptimal purification strategies (Figure [147]7d).
   This enhanced transfection efficiency and cytotoxicity underscore the
   critical importance of utilizing high‐purity circRNA in T cell
   engineering. The superior functional quality of WMC‐purified circRNA is
   essential for maintaining CAR‐T cell functionality and viability,
   ultimately contributing to the effectiveness of CAR‐T cell therapies.
   Our findings emphasize that circRNA purity is a key factor in
   optimizing T cell engineering processes for biomedical applications.

3. Discussion

   This study addresses the critical challenge of immunogenic impurities
   in synthetic circRNA, which can severely limit their therapeutic
   potential. Through comprehensive analysis, we identified key
   contaminants—such as dsRNA, RNA precursors, nicked RNA, and RNA
   phosphates—that significantly impact the immunogenicity and functional
   performance of circRNA. These impurities activate innate immune
   responses via RNA sensors, including PKR, OAS, TLR3, TLR7, TLR8, MDA5,
   and RIG‐I, triggering downstream signaling pathways such as NF‐κB, TNF,
   and TLR networks. Such immune activation not only reduces
   circRNA‐driven protein expression but also increases the risk of
   cellular dysfunction, highlighting the urgent need for effective
   purification methods to ensure the biosafety and functionality of
   circRNA therapeutics. Our findings demonstrate that even trace levels
   of these immunogenic impurities are sufficient to upregulate RNA
   sensors and inflammatory cytokines. As a result, achieving ultra‐pure
   circRNA is essential to minimize innate immune activation and ensure
   cell viability.

   To address these issues, the study developed a novel purification
   strategy involving a multi‐step process combining enzymatic treatments
   and cellulose‐based filtration. This method effectively removes the
   identified immunogenic impurities, thereby reducing the activation of
   immune responses and significantly improving the manufacturing yields
   and purity of circRNA. The use of cellulose chromatography was
   particularly effective in selectively binding dsRNA, further reducing
   immunogenicity. This novel purification approach resulted in highly
   pure circRNA with significantly reduced immunogenicity. The improved
   production efficiency achieved by this purification approach is
   approximately tenfold higher than traditional size‐based methods,
   highlighting its potential for large‐scale manufacturing.

   The findings of this study have significant implications for the
   production and application of circRNA in mRNA‐based cell programming.
   Our results emphasize the crucial importance of achieving high‐purity
   circRNA to maximize its functionality while minimizing cellular
   immunogenic responses. This is particularly beneficial for highly
   sensitive cells and challenging applications. Moreover, the integration
   of advanced purification techniques such as WMC chromatography with
   traditional methods could enhance the scalability and
   cost‐effectiveness of circRNA production. Continued investigation into
   the sources and impacts of RNA impurities will be essential to refine
   purification processes and improve the overall quality and safety of
   synthetic circRNA.

   Future efforts should prioritize refining and simplifying the
   multi‐step purification process to facilitate broader industrial
   application. Developing a user‐friendly, single‐step purification
   method would be ideal for large‐scale manufacturing, significantly
   enhancing its industrial applicability. To achieve this, minimizing
   undesired byproducts in crude products is essential. Optimizing T7
   polymerase to reduce dsRNA byproducts at the transcription source can
   alleviate downstream purification burdens, effectively addressing
   primary immunogenicity issues and improving both the safety and
   efficacy of the final product. Additionally, enhancing the efficiency
   of circRNA circularization will reduce linear RNA byproducts, further
   streamlining the purification process and improving overall
   manufacturing yields and purity. These strategic refinements are
   crucial for the scalable, cost‐effective production of high‐quality
   circRNA, which is vital for their successful biomedical and industrial
   applications.

   In conclusion, this study provides a comprehensive approach to
   mitigating the immunogenicity of synthetic circRNA, paving the way for
   their safe and effective use in mRNA‐based cell programming. The novel
   purification strategy developed here represents a significant
   advancement in circRNA research, offering a robust solution to one of
   the key challenges in the field. The stepwise purification strategy
   developed in this study effectively addresses the challenge of
   immunogenic impurities, paving the way for safer and more effective
   circRNA‐based therapeutics, especially for complex therapeutic
   applications. By minimizing the activation of innate immune responses,
   this approach has the potential to enhance the biomedical translation
   of circRNA technology.

4. Experimental Section

Cell Lines

   Human cervical carcinoma cell lines (HeLa), A549, and Human monocytic
   cell line THP‐1 were obtained from the Cell Resource Center, Peking
   Union Medical College (which is the headquarter of National Science &
   Technology Infrastructure‐National BioMedical Cell‐Line Resource,
   NSTI‐BMCR). Peripheral Blood Mononuclear Cells (PBMCs) were purchased
   from Oricells Biotech (Shanghai, China). Human fibroblast cells and
   Raji cells stably transduced with luciferase (CTCC‐0326‐Luc2) were
   purchased from MEISEN CELL (Zhejiang, China).

   HeLa and A549 cells were cultured at 37 °C and 5% CO[2] in Dulbecco's
   Modified Eagle's Medium (4500 mg L^−1 glucose) supplemented with 10%
   (v/v) heat‐inactivated fetal bovine serum (hi‐FCS, GIBCO) and 1% (v/v)
   penicillin/streptomycin (GIBCO). THP‐1 was maintained in RPMI 1640
   GlutaMAX medium supplemented with 10% (v/v) hi‐FCS (FCS) and a 1% (v/v)
   mixture of penicillin and streptomycin in a humidified incubator
   containing 5% CO[2] at 37 °C.

   Primary human T cells were isolated from PBMCs using the RosetteSep
   Human T cell Enrichment kit (Stem Cell Technologies) according to the
   manufacturer's protocol. Cryopreserved T cells were thawed and
   activated on D0 with ImmunoCult Human CD3/CD28/CD2 T Cell Activator
   (Stem Cell Technologies) in ImmunoCult‐XF T Cell Expansion Medium (Stem
   Cell Technologies). Mycoplasma testing was conducted regularly.

circRNA in Vitro Synthesis

   circRNA precursors were synthesized by T7 High Yield RNA Transcription
   Kit (Vazyme, China) according to the manufacturer's protocols. After
   transcription, reactions were treated with DNase I (Vazyme, China) for
   15 min. After DNase treatment, circRNA was incubated from reactions
   with 8 m LiCl. GTP was added to a final concentration of 2 mM along
   with a buffer including magnesium (50 mM Tris‐HCl, 10 mM MgCl[2], 1 mM
   DTT, pH = 7.5; New England Biolabs) into circRNA solution. Reactions
   were then heated to 55 °C for 8 min. After circularization, 20 U RNase
   R and 10 × RNase R buffer (Epicentre) was added according to the
   manufacturer's protocols and were heated to 37 °C for 20 min. After
   RNase R digestion, circRNA was collected using LiCl precipitation. Then
   added 1 U µg^−1 Quick CIP (New England Biolabs) and 10 × rCurSmart into
   circRNA and heated 37 °C for 20 min.

FPLC Based Purification of circRNA

   Two types of cellulose were used for circRNA purification:
   microcrystalline cellulose (MCC) and Natural Wood‐Derived Macroporous
   Cellulose (WMC).

   For the MCC‐FPLC method, 450 mg of MCC was added to a 1 mL column,
   which was then connected to the FPLC system and washed with
   chromatography buffer (10 mM HEPES, 0.1 mm EDTA, 125 mm NaCl, and 25%
   or 30% (v/v) isopropanol) at a flow rate of 1 mL min^−1 for 20 column
   volumes. Subsequently, 100 µg of circRNA was loaded into the column.
   After purification, the isopropanol‐containing eluent was precipitated
   with LiCl for recovery.

   For WMC‐FPLC, the cellulose was synthesized as described in.^[ [148]^34
   ^] 250 mg of WMC was added to a 1 mL column, and the WMC‐FPLC
   purification followed the same process using a different chromatography
   buffer (10 mm HEPES, 0.1 mm EDTA, 125 mm NaCl, and 16% (v/v) ethanol,
   pH 7.2).

Spin Column Based Purification of circRNA

   5.3 mg of WMC was filled into a spin column and washed using 500 µL
   elution buffer (10 mm HEPES, 0.1 mm EDTA, 125 mm NaCl, and 16% (v/v)
   ethanol, pH = 7.2) through vigorous shaking for 5 min. After 7000 rpm
   centrifugation for 10 s, the filtrate was discarded. Repeat the washing
   process several times, 10 µg of circRNA sample with 200 µL elution
   buffer was loaded into the column, and the mixture was shaken
   vigorously for 6 min. The unbound single‐stranded mRNA in the effluent
   was collected and separated with dsRNA by 14000 × g centrifugation for
   1 min. For recovery, circRNA was precipitated with LiCl.

SEC Purification of circRNA

   circRNA was run through a 7.8 × 300 mm size exclusion column (Sepax
   Technologies; part number: 215950–7830) on AKTA FPLC (Cytiva). RNA was
   run in RNase‐free Potassium Phosphate buffer (10 mm potassium
   phosphate, 1 mm EDTA, pH = 6) at a flow rate of 0.4 mL min^−1. For
   recovery, circRNA was precipitated with LiCl.

Capillary Electrophoresis

   RNA was tested using the 5200 Fragment Analyzer system (Agilent) with
   the Fragment Analyzer RNA Kits (DNF‐471‐0500) according to the
   manufacturer's instructions.

HPLC

   circRNAs were analyzed by an ion‐pair reversed‐phase (IP‐RP) HPLC with
   a 1260 system from Agilent Technologies. The analysis was conducted on
   a 250 × 4 mm reverse‐phase LC column with a particle size of 5 µm and a
   pore size of 4000 Å (Agilent Technologies: PLRP‐S 4000Å). A linear
   gradient was employed with buffer A (100 mm hexyl ammonium acetate
   [HAA] in 10% acetonitrile) and buffer B (100 mm HAA in 50%
   acetonitrile) at a flow rate of 0.4 mL min^−1. The column oven
   temperature was maintained at 60 °C, and RNA was detected by UV
   absorbance at 260 nm.

Lateral Flow Strip Assay

   The test strip preparation followed the approach outlined in a previous
   report (31). 100 ng of circRNA was mixed with a 20% BSA (10 mm
   Tris‐HCl, pH = 8.0). The final total volume was 50 µL. The mixture was
   then added to the sample area of the test strip. After 15 min, the test
   strip was imaged on the ChemiDoc MP Imaging. The exposure time was set
   to 0.4 s.

Cell Transfection

   For HeLa cells, 400 ng of RNA was transfected into 4 × 10^4
   cells/200 µL per well of a 48‐well plate using Lipofectamine 3000
   (Invitrogen) according to the manufacturer's instructions. To obtain
   PMA‐differentiated THP‐1 cells, THP‐1 cells were seeded at a density of
   2.5 × 10^5 cells/500 µL per well in a 12‐well culture plate and
   stimulated with 100 ng mL^−1 of PMA for 48 h. A549 cells were seeded at
   a density of 2 × 10^5 cells/500 µL per well in a 12‐well culture plate
   24 h prior to transfection. For fibroblasts, 10^4 cells/200 µL per well
   24 before transfections. A total of 1 µg or 200 ng of RNA, or 40 ng of
   HMW or intron fraction was transfected into A549, fibroblasts, and
   THP‐1 cells. All the cells were transfected using Lipofectamine 3000
   (Invitrogen) according to the manufacturer's instructions.

   For PBMCs, 1 µg of RNA was added to 2 × 10^5 cells re‐suspended in
   10 µL of T buffer R from the Neon Transfection System 10 µL Kit
   (Invitrogen). Electroporation was performed at 1600 V, 10 ms, 3 pulses
   settings using a Neon electroporation device (Invitrogen). Upon
   electroporation, the cells were seeded on the recovery plate with T
   cell media and 100 IU mL^−1 of IL‐2.

circRNA Reprogramming

   Fibroblasts were plated on dishes coated with matrigel.
   (Corning)following the manufacturer's instructions, at densities of
   6000 cells per well in a 6‐well dish format. The cells were cultured in
   reproTESR (Stemcell)″ supplemented with 200 ng mL^−1 B18R (novo
   protein). The medium was replaced daily, both before and after circRNA
   transfections. B18R supplementation was discontinued the day after the
   final transfection. Different rounds of: 24‐h transfections with
   circRNA were performed as previously described^[ [149]^35 ^] using
   LipoMessengerMAX (Invitrogen)according to the manufacturer's
   instructions. Each transfection delivered 800 ng of' circRNA per well,
   containing Oct4, Sox2, Klf4, c‐Myc, Lin28A, and Nanog at·a molecular
   ratio of 3:1:1:1:1:1.

Cellular Fluorescence and Flow Cytometry

   24 h after RNA transfection, cells were imaged on a fluorescence
   microscope (OLYMPUS CKX53). For eGFP detection, cells were washed three
   times with PBS and resuspended in 100 µL of PBS before analysis on a
   flow cytometer (ACEA NovoCyte, Agilent). For surface staining, T cells
   were washed twice in FACS buffer (2% FBS in PBS) before staining with
   fluorescent‐labeled antibodies. The cells were resuspended in 100 µL of
   antibody solutions and incubated at 4 °C for 30 min. After incubation,
   the cells were washed twice and resuspended in 100 µL of PBS before
   analysis on a flow cytometer (ACEA NovoCyte, Agilent). Data were
   analyzed using FlowJo software version 10.8. Transfection efficiency
   was quantified as the percentage of GFP‐positive cells relative to the
   total cell population. The expression level of GFP mRNA was represented
   by the MFI of fluorescein isothiocyanate (FITC) in positive cells. For
   CAR expression level, cells were prepared in staining buffer (PBS
   supplemented with 2% FBS) and subjected to fluorochrome‐conjugated
   antibody (APC‐Labeled Monoclonal Anti‐FMC63 Antibody, Acro, FM3‐AY54A1)
   staining for surface markers, following standard protocols.

Cytotoxicity Studies

   The cytotoxicity studies were conducted according to.^[ [150]^39 ^]
   Raji–luc target cells were pelleted with CAR‐T cell products in 96‐well
   plates at effector: target (CAR^+ T‐cell: Raji–luc) ratios of 10:1,
   5:1, 2.5:1, 1:1, and 0:1 for 20 h. 1% triton X‐100 was used as a
   positive control. 1 × 10^5 CAR^+ T cells and 1 × 10^4 target cells were
   used as the highest effector‐to‐target ratio in a total volume of
   100 µL RPMI supplemented with 10% FBS. After 20 h, CAR‐T cells and
   Raji–luc co‐cultures were lysed and the luminescence signal of the
   solution was quantified using a SpectraMAX M5 fluorescence microplate
   reader (Molecular Device). Percent cytotoxicity was calculated as
   1 − (sample − ctrl[triton X‐100])/(ctrl[0:1] − ctrl[triton X‐100])%.

RNA Isolation and RT‐qPCR

   Total RNAs were extracted with FastPure Cell/Tissue Total RNA Isolation
   Kit V2 (Vazyme, China) according to the manufacturer's protocol. 0.1 mg
   of total RNAs were served as the template for reverse transcription by
   using PrimeScript II 1st Strand cDNA Synthesis Kit (TAKARA) according
   to the manufacturer's protocols. After reverse transcription, cDNAs
   were transcribed with Taq Pro Universal SYBR qPCR Master Mix (Vazyme,
   China) according to the manufacturer's protocol. Primer pairs were
   selected using the Primer Blast designing tool
   ([151]https://www.ncbi.nlm.nih.gov/tools/primer‐blast/). Relative
   expression of mRNA abundance was determined from three independent
   experiments and was normalized by housekeeping gene B2 m or GAPDH.

RNA Sequencing

   The total RNA of each sample was isolated and purified using Trizol
   reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer's
   procedure. The RNA amount and purity of each sample were quantified
   using NanoDrop 2000 (NanoDrop, Wilmington, DE, USA). The RNA integrity
   was assessed by Agilent 2100 with an RIN number >7.0. Then, the mRNA of
   each sample was enriched using oligo(dT) magnetic beads. RNA‐seq
   libraries with an average insert size of 350 bp were constructed using
   AHTS Universal V8 RNA‐seq Library Prep Kit for Illumina (Vazyme, China)
   following the manufacturer's instructions, and then sequenced on
   SURFSeq 5000 platform (GeneMind Biosciences LTD., China) with PE150
   model. SOAPnuke^[ [152]^40 ^] was adopted to remove the reads with an
   adaptor, ploy‐N, and low‐quality bases (‐n 0.01 ‐l 15 ‐q 0.5 –ada_trim
   –minReadLen 75 –polyX 50). The Cleaned reads were then aligned to the
   human reference genome (GRCh38) with Histat2.^[ [153]^41 ^] The mapped
   reads were delivered to StringTie to perform expression
   quantification.^[ [154]^42 ^] Differentially expressed
   genes/transcripts (DEGs/DETs) were identified with p.value < 0.05 and
   |log2(fold change) | ≥1 by the R package DESeq2.^[ [155]^43 ^] The KEGG
   and GO enrichment of DEGs was conducted with the R package
   clusterProfiler^[ [156]^44 ^] with p.value < 0.05 as the threshold for
   significantly enriched DEGs.

Statistics Analysis

   Data were presented as mean ± standard deviation (SD). The specific
   sample size (n) for each experiment is reported in the figure legends.
   Statistical significance was assessed using one‐way ANOVA, with a
   significance level set at p < 0.05. Any assumptions required for the
   chosen statistical analysis were checked before testing. All analyses
   were conducted using GraphPad Prism 9.0.

Conflict of Interest

   The authors declare no conflict of interest.

Supporting information

   Supporting Information
   [157]ADVS-12-2416629-s001.docx^ (2.4MB, docx)

Acknowledgements