A convenient method to generate and maintain poly(A)-encoding DNA sequences required for in vitro transcription of mRNA

Use of mRNA to encode therapeutic, cell-modifying or immunogenic proteins has recently gained popularity [1]. Valuable features of mRNA for such application are: 1) a cytoplasmic site of action, unlike DNA, which simplifies delivery when formulated in synthetic nonviral vectors; 2) RNA is shortlived in vivo to make control of dose easier to achieve, and 3) mRNA does not recombine with DNA, and resultant mutation of the host genome is avoided.

For large-scale preparation of mRNAs in vitro, sequences are typically generated using T7 RNA polymerase. Products of the transcription reaction usually have a 5′ CAP or internal ribosomal entry site (IRES), 5′ untranslated region (5′UTR), protein-coding open reading frame (ORF), 3′UTR and poly(A) sequences at the 3′ end. The number of A residues at the 3′ end of mRNA is variable and has been reported to range between 67 to 96 nt in actively translated mRNAs of mammalian cells [2]. Ensuring an optimal length of the poly(A) tail is vital for efficient translation and shortening of these sequences leads to decay of mRNA (reviewed in [3,4]). Multiple A residues may be added to synthetic RNA using a poly(A) polymerase from Escherichia coli (EPAP) [5] or during transcription from an encoding template [6,7]. When using EPAP to add A residues to the 3′ end of mRNA, achieving consistency of the tail lengths may be difficult [6,7]. A complication of using poly(A:T)-containing circular plasmids is that the homopolymeric A:T base pairs in these vectors are unstable and prone to shortening during replication in bacteria [6,8]. Use of a linear plasmid vector was recently employed to increase stability of longer poly(A:T) tracts [6].

To improve on generating poly(A)-encoding templates from widely used circular plasmid DNA, we developed template DNA that may be lengthened using standard procedures of restriction digestion with ligation and propagation in bacteria. The starting DNA used to engineer the plasmid DNA sequences is depicted in Figure 1A. An mRNA-generating plasmid with 5′UTR comprising the entire IRES of hepatitis C virus (HCV) [9] and 3′UTR derived from human β-globin was generated in house. Insertion of a gene of interest, which was Firefly Luciferase (FLuc) in the example used here, was achieved using Gibson assembly [10]. The sequence of the poly(A:T)-encoding region comprised 13 A:T base pairs with flanking FaqI and BveI type IIS restriction sites. The BveI site was positioned at the 5′ end and the FaqI site at the 3′ end of the poly(A:T)-encoding sequence. These type IIS restriction enzymes have unique cleavage sites in the poly(A:T) flanking regions. Orientation of the recognition sites was such that the enzymes each cleaved asymmetrically within the poly(A:T) sequence to generate complementary sticky ends comprising four T or A overhangs. In addition, a SapI restriction site was placed immediately 3′ of the poly(A:T) tracts, and this enzyme was used to linearize the plasmid to generate a template for in vitro transcription. A SapI site that naturally occurs within the FLuc ORF was abolished using site-directed mutagenesis while retaining the amino acid sequence of the reporter enzyme. The first step of the process to lengthen the A:T sequence entailed two independent double digestions of the plasmid (Figure 1B). After cleavage with BveI and ScaI or FaqI and ScaI according to procedures recommended by the supplier of the enzymes (New England Biolabs, MA, USA), fragments that each contained the longer poly(A:T) sequences were purified. A replicative plasmid was reconstituted after ligation (T4 DNA ligase, ThermoFisher, MA, USA) and propagated in chemically competent bacteria of the XL1-Blue strain (Agilent Technologies, CA, USA). ScaI has a unique recognition site within the ampicillin resistance gene and was selected to minimize generation of unintended plasmid clones. The plasmid formed following this initial digestion and ligation should thus have contained a poly(A:T) sequence of 18 bp. Restriction digestion and ligation of the plasmids were repeated and the poly(A:T) tract lengthened with each successive cycle. Theoretically the length of the poly(A:T) sequence would be increased to 28, 48, 88 and then 168 bp for the second to fifth cloning cycles, respectively. Precise measurement of the number of A:T base pairs was not possible using standard Sanger sequencing (Figure 1C) and the length of the homo tract was therefore estimated from the length of restriction fragments that encompassed the region of interest (Figure 1D). Our observations indicated a maximum length of the poly(A:T) tract of 80 to 100 bp was achieved after six cycles of the repeated cloning process. This is similar to naturally occurring poly(A) in mRNA and should be suitable for intracellular translation [2]. Information available online indicates that this length of the poly(A)-encoding sequences also exceeds what can typically be guaranteed by commercial suppliers of synthetic duplex DNA. Linear templates derived from intermediates of the sequential steps of the cloning reactions, which lacked FLuc sequences, were used to generate mRNA using T7 RNA polymerase (TranscriptAid T7 High Yield Transcription Kit, ThermoFisher). Electrophoretic resolution of the mRNA confirmed increasing lengths of the transcripts, which correlated with the lengthening poly(A)-containing 3′ ends of the mRNA (Figure 1E).

To evaluate whether the poly(A)-containing mRNA sequences were functional, FLuc-encoding plasmid DNA was also linearized with SapI and transcribed in vitro using T7 RNA polymerase. Assessment of translation from the template in vivo was determined using an institutionally approved protocol (AESC 2018/21/5C) for transfection of murine hepatocytes using hydrodynamic injection [11]. Expression of FLuc in the liver was measured using standard bioluminescence imaging with an IVIS Kinetic Bioluminescence imager (PerkinElmer, MA, USA). Representative data obtained 3 h after administration of 5 μg of mRNA are shown in Figure 1F. After an exposure of 60 s, 1.045 × 10⁷ photons were detected in the region of interest over the liver of the mouse shown in Figure 1F. This value is comparable to that observed after similar administration of commercially available FLuc-encoding mRNA (Trilink, CA, USA). These data verify that the poly(A) tail of the in vitro transcribed transcript was functional in vivo.

The simple method described here is convenient for propagation of plasmids with transcripts that may be used to serve as templates for preparation of efficiently translated mRNA. Moreover, should shortening of the sequence occur, reconstitution of the length of the poly(A) tract could be conveniently achieved.