*3.1. Optimization Strategy for Splenic DCs*

Microfluidic technology has made it possible to reproducibly produce relatively small and uniform LNPs and has recently been adopted as a major production method [42]. The rapid and reproducible mixing of two liquids (an alcohol solution of lipids and a buffer solution containing RNAs) is achieved by introducing them at a high flow rate into a microfluidic device equipped with a micromixer. While most RNA-loaded LNPs that are formed by microfluidic technology are less than 100 nm in size [43–46], we recently reported that the addition of a salt (e.g., NaCl) to the RNA-containing buffer significantly contributed to the formation of LNPs with sizes in excess of 100 nm based on the Derjaguin– Landau–Verwey–Overbeek (DLVO) theory [39]. The addition of a salt reduces electrostatic repulsion and promotes fusion between the initially produced liposome-like cationic particles, resulting in the formation of larger LNPs in a salt concentration-dependent manner. Both cellular uptake and the functional delivery of siRNA and mRNA in splenic DCs were significantly higher when larger LNPs were used compared to the smaller counterparts with a consistent lipid composition. This can be explained by the fact that macropinocytosis, which is a unique pathway characterized by the nonspecific internalization of large amounts of extracellular fluid, is constitutively active in immature DCs and, therefore, relatively larger-sized particles (e.g., 200 nm or higher) would be beneficial for targeting DCs [47,48]. Although the finding suggests that an increase in LNP size is a significant factor for the efficient delivery of RNAs to splenic DCs, the optimal ranges of the size and other factors, including lipid composition, have not yet been clarified.

In this study, we first conducted 2 steps of DOE for screening and optimizing both synthetic conditions (NaCl concentration) and the lipid composition of LNPs for the delivery of mRNA to splenic DCs. The experimental scheme is represented in Figure 1. The physicochemical properties and cellular uptake of the LNPs, which would be mediator

variables, were also measured in an attempt to understand how these factors contribute to the functional delivery (final output). Nluc mRNA was used to quantify the transgene expression level. The long intracellular half-life of the Nluc protein (>6 h) attenuates the potential effects of protein degradation during the lengthy (at least several hours) experimental process from sacrifice to cell sorting and would be suitable for the measurement of enzymatic activity in the sorted cells. The bright signal from Nluc was also suitable for quantitatively detecting transgene expression levels in the limited number of sorted cells. DiO-labeled LNPs were used to assess cellular uptake. Three types of APCs, including DCs, macrophages (Mϕ), and B cells, were used in the analysis.

**Figure 1.** Schematic illustration of the experimental method for DOE. (**A**) Parameters and responses in manufacturing LNPs. Both parameters and responses examined in this study were expressed in bold. (**B**) Method for screening of LNPs in vivo. Cellular uptake, NHC class II expression, and Nluc expression in splenic DCs, Mϕ, and B cells after intravenous injection of DiO-labeled mNLuc-loaded LNPs selected by DOE were measured to identify the optimal formulation.

Adjuvant effects associated with type I interferon (IFN) (IFN-I) stimulation have been suggested to lead to superior acquired immune responses, and pathways that activate IFN-I expression have been identified, including the TLRs, RLRs, and stimulator of interferon genes (STING) pathways [12]. Mouse TLR7, which recognizes single-stranded RNA (ssRNA) and activates the adaptor protein myeloid differentiation primary response 88 (MyD88), leads to the expression of a suite of inflammatory cytokines, including IFN-I [49]. It is generally thought that regular uridine-containing mRNA, which is an ssRNA, stimulates the innate immune system via TLR7. The cytoplasmic RNA sensor RIG-I binds to double-stranded RNA (dsRNA) and induces the expression of inflammatory cytokines such as IFNβ through the activation of the adapter protein mitochondrial antiviral signaling protein (MAVS) [50]. Trace amounts of dsRNA, a byproduct of the in vitro transcription of mRNA, contribute to IFNβ production via the RIG-I/MAVS pathway [51,52]. In this study, we used regular uridine-containing mRNAs, which were obtained from Trilink BioTechnologies, which are thought to induce IFN-I expression through stimulation of the innate immune system via the RIG-I/MAVS pathway and TLR7. IFN-I is known to drive a distinctive DC maturation program, including the continuous upregulation of MHC-II and antigen processing [53], and to be highly expressed by DCs (especially by plasmacytoid DCs that highly and constitutively express IFN response factor-7) [54]. Therefore, the efficient introduction of immunostimulatory mRNAs into DCs results in the efficient maturation

of DCs and, for dendritic cells, the expression level of MHC class II (I-A/I-E), one of the maturation markers, was also quantified in the screening.
