Robust genome and cell engineering via in vitro and in situ circularized RNAs

Abstract

Circularization can improve RNA persistence, yet simple and scalable approaches to achieve this are lacking. Here we report two methods that facilitate the pursuit of circular RNAs (cRNAs): cRNAs developed via in vitro circularization using group II introns, and cRNAs developed via in-cell circularization by the ubiquitously expressed RtcB protein. We also report simple purification protocols that enable high cRNA yields (40-75%) while maintaining low immune responses. These methods and protocols facilitate a broad range of applications in stem cell engineering as well as robust genome and epigenome targeting via zinc finger proteins and CRISPR-Cas9. Notably, cRNAs bearing the encephalomyocarditis internal ribosome entry enabled robust expression and persistence compared with linear capped RNAs in cardiomyocytes and neurons, which highlights the utility of cRNAs in these non-dividing cells. We also describe genome targeting via deimmunized Cas9 delivered as cRNA and a long-range multiplexed protein engineering methodology for the combinatorial screening of deimmunized protein variants that enables compatibility between persistence of expression and immunogenicity in cRNA-delivered proteins. The cRNA toolset will aid research and the development of therapeutics.

Fig. 1 |. Engineering ocRNAs and icRNAs.

a, Schematic describing the production of ocRNAs. These are generated via IVT of linear RNAs that bear twister ribozyme + permuted group II intron–exon sequence flanked IRES coupled to an mRNA of interest. Once transcribed, the flanking twister ribozymes rapidly self-cleave, enabling hybridization of the complementary ligation stems to one another. Then, autocatalytic intron splicing occurs, releasing the introns and ligating the spliced ends together. b, Sanger sequencing trace mapping the junction site formed upon ligation. c, Tapestation of ocRNA before and after RNaseR treatment. d, Circularization efficiencies quantified by tapestation analysis. Values represented as mean ± s.e.m. (n = 11). e, Yields from cellulose dsRNA purification across differing length RNA constructs. f, Schematic describing the production of icRNAs. These are generated via IVT of linear RNAs that bear a twister ribozyme flanked IRES coupled to an mRNA of interest. Once transcribed, the flanking twister ribozymes rapidly self-cleave, enabling hybridization of the complementary ligation stems to one another, and upon delivery into cells, these linear RNAs are then circularized in situ by the ubiquitous RNA ligase RtcB. g, Sanger sequencing trace mapping the junction site formed upon ligation. h, HEK293Ts were transfected with icRNA or in vitro pre-circularized icRNA, and RNA was isolated at 6 h, 24 h and 48 h. RT-PCR was performed and the ratio of the icRNA band to pre-circularized icRNA band was plotted to evaluate in situ circularization efficiencies. Values represented as mean ± s.e.m. (n = 3). i, HEK293Ts were transfected with icRNA and RNA was isolated at 6 h, 24 h and 48 h and RNAseq performed. cRNA counts relative to total RNA are shown. Values represented as mean ± s.e.m. (n = 3).

Fig 2 |

a, HEK293Ts were transfected with icRNAs containing various IRES sequences and GFP intensity was quantified by flow cytometry. cRNAs containing the EMCV IRES (blue bar) were selected for further optimization. Values represented as mean ± s.e.m. (n = 3). b, HEK293Ts were transfected with icRNAs containing the EMCV IRES coupled with various 3′ UTRs and poly(A) stretches (blue bars) and GFP intensity was quantified by flow cytometry. Addition of a WPRE and a poly(A) stretch substantially improved protein translation, and icRNAs bearing the EMCV IRES were used for all subsequent studies. These designs were also compared with capped linear N¹-methylpseudouridine-5′-triphosphate (m1Ψ) RNA (red bar). Values represented as mean ± s.e.m. (n = 3). c, A549swere transfected with various constructs encoding GFP and RNA isolated at 6 h, 24 h and 48 h. Immune markers RIG-I, IFNB and IL6 were quantified by RT-qPCR for each sample relative to GAPDH. Values represented as mean ± s.e.m. (n = 3). Shown below are detailed methods of synthesis and purification for each construct. d, A549s were transfected with the same constructs encoding GFP, and RNA was isolated at 6 h, 24 h and 48 h. Cell viability was quantified via CCK-8. Absorbance at 450 nm was measured for samples on day 0, day 1 and day 2. Data were normalized within each sample to day 0 values. Values represented as mean ± s.e.m. (n = 3). e, HEK293T cells were transfected with circular GFP icRNA containing encephalomyocarditis IRES with WPRE and 50 nt poly(A) stretch or 5′ capped linear RNA and the GFP mRNA amount was measured over time (left axis). Values were normalized to the amount at the 6 h time point for each respective group (n = 3, P = 0.0072 for day 1, P = 0.0015 for day 2, P = 0.00086 for day 3, P = 0.0037 for day 4 and P = 0.000531 for day 5; t-test, two-tailed). The ratio of icRNA GFP mRNA compared with linear RNA for each day is plotted (right axis). The increase in value over time illustrates improved persistence of icRNA.

Fig. 3 |. Assessing persistence and activity of ocRNAs and icRNAs.

a, Indicated cell types were transfected with linear m1Ψ and icRNA. GFP intensity was quantified on day 1 by flow cytometry relative to icRNA. (Cardiomyocytes and neurons are represented by calculated total cell fluorescence (CTCF) image data.) Values represented as mean ± s.e.m. (n = 3). b, Left: post differentiation of stem cells into neurons, icRNAs or linear m1Ψ RNAs were transfected into cells and images were taken over 10 days. CTCF mCherry expression over time was plotted for icRNA and linear m1Ψ RNA. c, Left: post differentiation of stem cells into cardiomyocytes, ocRNAs, icRNAs or linear m1Ψ RNAs were transfected into cells and images were taken over 30 days. Middle: CTCF GFP expression over time was plotted for icRNA and linear m1Ψ RNA. Right: relative GFP RNA quantified with RT-qPCR after 30 days. Values represented as mean (n = 2). Bottom: representative images are shown illustrating icRNA and ocRNA persistence. d, Left: schematic of a one-time transfection of 500 μg icRNA encoding NeuroD1-P2A-GFP onto stem cells (H1). Middle: day 7 TUBB3 immunostaining of differentiated neurons. Right: neural markers MAP2, TUBB3, BRN2 and vGLUT2 were quantified by RT-qPCR for each sample relative to GAPDH. Values for GFP are represented as mean (n = 2) and values for NeuroD1-P2A-GFP are represented as mean ± s.e.m. (n = 3).

Fig. 4 |. Application of icRNAs and ocRNAs to ZF-mediated genome and epigenome targeting.

a, Editing efficiency of circular icRNA or circularization defective icdRNA ZFNs targeting a stably integrated GFP gene or the endogenous CCR5 gene in HEK293T cells is plotted. Values represented as mean ± s.e.m. (n = 3). b, Target locations of ZFs are indicated alongside forward (sense) and reverse (antisense) strand binding. Repression efficiency of ZF-KRAB proteins produced by circular icRNA in HeLa cells is plotted on the left as hPCSK9 expression fold change relative to the circular GFP icRNA quantified with RT-qPCR after 48 h. The black dotted line indicates the measure for successful repression of hPCSK9 by a human ZF-KRAB protein. The green dotted line indicates hPCSK9 levels in the GFP control. c, Transient hPCKS9 repression efficiency produced by a 3A3L and KRAB fusion with ZF10 delivered as ocRNA in HeLa cells. hPCSK9 levels are quantified with RT-qPCR at each time point. Values represented as mean ± s.e.m. (n = 3).

Fig. 5 |. LORAX protein engineering methodology to screen progressively deimmunized Cas9 variants.

Left: library design. Low-frequency SNPs that have a limited effect on Cas9 function were identified and immunogenicity was evaluated in silico using the netMHC epitope prediction software to identify candidate mutations. This analysis was performed for many Cas9 orthologues. Mutations were generated such that 2 bp was changed to account for nanopore sequencing accuracy. A library was then generated by fusion PCR of blocks containing WT and mutations at specific epitopes. Location of epitopes in SpCas9 that were combinatorially mutated and screened is shown. Right: library screen. The screen was performed by transducing HeLa cells with a lentiviral library containing the Cas9 variants and a guide that cuts the HPRT1 gene. HPRT1 knockout produces resistance to 6-TG. After 2 weeks, DNA is extracted from surviving cells and Cas9 variant sequences are PCR amplified from the genomic DNA and nanopore sequenced. High accuracy of variant identification is possible owing to the use of 2 bp mutations for each amino acid change. Post-screen library element frequencies across two independent replicates are shown. Replicate correlation was calculated excluding the over-represented WT sequence.

Fig. 6 |. Validation of LORAX screen identified Cas9 variants for deimmunization, and genome and epigenome targeting via delivery as icRNAs and ocRNAs.

a, Network reconstruction connecting Cas9 variants with similar mutational patterns. Node colours indicate the number of deimmunized epitopes (dark blue < 3, light blue = 3, white = 4, yellow = 5, pink > 5). Circles in red represent tested variants and labelled with their respective names. b, HEK293T bearing a GFP coding sequence disrupted by the insertion of a stop codon and a 68 bp genomic fragment of the AAVS1 locus were used as a reporter line. WT or Cas9 variants, an sgRNA targeting the AAVS1 locus and a donor plasmid capable of restoring GFP function via HDR were transfected into these cells and flow cytometry was performed on day 3. Relative quantification of GFP expression restoration by HDR is plotted. The number in parentheses represents the number of mutations in the variant. Values represented as mean ± s.e.m. (n = 3). c, T2 cells were pulsed with WT and variant peptides, cultured with PBMCs, and an ELISpot assay was performed to assess PBMC IFNγ secretion to WT and variant peptides. The number of spot-forming colonies for each peptide is plotted (n = 3, mean ± s.e.m., *P < 0.05, **P < 0.01, unpaired t-test, two-tailed). Red letters in the peptide sequences represent the mutated amino acid. d, RNA encoding for Cas9 WT or variant V4 was electroporated into PBMCs to assess the whole protein immunogenicity. ELISpot assay was performed to assess PBMC IFNγ secretion to WT and variant protein. The number of spot-forming colonies for each peptide is plotted (n = 3, mean ± s.e.m., ****P < 0.0001, unpaired t-test, two-tailed). e, Circular icRNA for Cas9 WT or variant V4, along with an sgRNA targeting the AAVS1 locus, was introduced into HEK293T and K562 cells. Editing efficiency at the AAVS1 locus in the two cell lines is plotted. Values represented as mean ± s.e.m. (n = 3). f, Circular icRNA and ocRNA for CRISPRoff WT or variant V4, along with an sgRNA targeting the B2M gene, were introduced into HEK293T cells. B2M gene repression of CRISPRoff constructs in the presence or absence of sgRNA is plotted. Values represented as mean ± s.e.m. (n = 3).