**2. Background**

In 1990, Felgner and colleagues published, in Science [6], their demonstration that so-called "naked DNA", that is, plasmid DNA that was not formulated in transfecting agents, could be directly injected into muscle with resultant expression of the encoded protein by myocytes. The observation was important because up until then, significant effort had been devoted to formulations to deliver DNA in vivo, and many such compounds were used for in vitro transfection. The surprising simplicity of the approach generated significant interest, and when it was soon shown (in 1993) that plasmid DNA coding for a conserved internal influenza protein could generate protection in a pre-clinical mouse model against influenza challenge with a very different influenza strain than the strain from which the protein antigen was sequenced [7], many groups began developing plasmid DNA for vaccines, cancer immunotherapies, and immune interventions for autoimmune and allergic diseases.

The same 1990 publication also demonstrated that naked RNA could similarly result in the in vivo expression of encoded protein. However, more attention focused on utilizing plasmid DNA, rather than mRNA, likely because of concerns about the instability of mRNA. In 1992, Bloom and colleagues [8] demonstrated the efficacy of mRNA to express protein in vivo by showing that mRNA encoding a hormone could correct a disease following direct injection into rat brains. In the same year (1993) that the first demonstration of the ability of DNA plasmid to protect mice from heterosubtypic challenge with influenza was published [7], liposome-formulated mRNA was also shown to generate influenza-specific cytolytic T cells in mice [9] (although protection from infectious challenge, as was shown for plasmid DNA, was not tested, perhaps explaining part of the difference in excitement about the technologies). Nevertheless, for both entities, a key issue was how to optimally deliver the DNA plasmid or the mRNA into the desired cells, either for optimal expression of the desired therapeutic protein as a drug or for gene therapy (to supply a missing or defective protein), or to generate the desired immune response against the protein if it were an antigen. For gene therapy, the encoded protein needs to not stimulate an immune response. For a vaccine, which cell produces the protein encoded by the mRNA or the plasmid DNA can be a key issue because, although for antibodies [10] where the protein would likely need to be secreted, for cellular immune responses of the Cytolytic T Lymphocyte variety [11], the type of cell producing the protein (and hence the cell type transduced by the plasmid DNA or the mRNA) is relevant, as is discussed later.

Why was there relatively less interest in mRNA compared to plasmid DNA as a platform technology for over a decade? What led to the recent explosion of interest and progress for mRNA? The transient nature of messenger RNA, possibly an asset for the process whereby organisms control the production of desired proteins, is due to RNAses that are widely present [12]. This instability of mRNA has been a significant reason for the lack of interest in mRNA as a drug. In addition, RNA has long been known to be an immunologically active molecule. For example, poly (I:C) (polyinosine-polycytidylic acid) is a synthetic analog of dsRNA (double-stranded RNA) that is an agonist of TLR3 and has long been used as an immunostimulatory mimic of viral infection and tested as an adjuvant to increase immune responses for experimental vaccines [13–15]. mRNA has a number of immunostimulatory mechanisms, which may be useful—or detrimental—for mRNA used for vaccines or cancer immunotherapeutics (discussed below). However, these properties contributed to the lower degree of interest in mRNA versus plasmid DNA for gene therapy applications when provision of a missing or defective protein with no immune responses against either the protein or the vector delivery system was the goal.

Two developments were important for changing the perception and reality of mRNA. These were the demonstration by Weissman and Kariko that the use of modified nucleosides made in vitro-transcribed mRNA less immunogenic [16]. In follow-up work, they showed that using pseudouridine instead of uridine resulted in mRNA that was more stable and had increased translational capacity [17]. This use of modified nucleosides thus addressed key issues for mRNA—stability of the mRNA, increased production of the encoded protein, and some decrease of the innate immunogenicity. Additional work explored the use of other nucleosides, such as substituting 5-methylcytidine for cytidine with further improvement [2,11].
