Microcondensate‐Mediated Intracellular Infusion of mRNA Across the Plasma Membrane

Abstract

This study presents a strategy for developing micrometer‐sized coacervates comprised of cationic intracellular delivery peptide FcB(L17E)₃ and nucleic acids (termed “microcondensates”) as a lipid nanoparticle (LNP)‐free platform for intracellular messenger RNA (mRNA) delivery. The feasibility of this approach is demonstrated both in vitro and in vivo. This approach is based on our previous finding that FcB(L17E)₃ forms a microcondensate through interactions with IgG labeled with a negatively charged fluorescent dye. Similarly, the negative charge of nucleic acids causes siRNAs to form microcondensates with FcB(L17E)₃, resulting in the facile infusion of the siRNAs through the plasma membrane with 40% gene knockdown efficiency. mRNAs form aggregates with FcB(L17E)₃ due to their longer chains, which prevents efficient delivery. Remarkably, adding short single‐stranded DNA (ssDNA) to mRNA enables microcondensate formation with FcB(L17E)₃, facilitating protein expression. Additionally, this microcondensate system successfully facilitated the intracellular delivery of plasmid DNA (pDNA) by optimizing the charge ratio between FcB(L17E)₃ and nucleic acids. Finally, the subcutaneous injection of microcondensates into mice successfully induced protein expression in vivo. The achievement of in vivo mRNA delivery and protein expression using microcondensates may offer a strategy for nucleic acid delivery with potential applicability in cancer immunotherapy and mRNA‐based vaccination.

The delivery of mRNA without lipid nanoparticles remains challenging. FcB(L17E)₃‐driven microcondensates with nucleic acids, aided by short ssDNA, enable coacervate formation, actin‐dependent cytosolic infusion, and in vivo protein expression, offering a potential platform for next‐generation mRNA therapeutics and vaccination.

Introduction

Messenger RNA (mRNA) therapeutics have emerged as a groundbreaking approach in modern medicine, offering the potential to treat diseases previously considered untreatable.^{[ 1 ]} Their ability to transiently express proteins without the risk of genomic integration, coupled with scalable and efficient manufacturing processes, has established mRNA as promising alternative to DNA‐based therapies for diverse applications, including vaccinology, allergy desensitization, gene therapy, and cancer immunotherapy.^{[ 2 ]} Notably, lipid nanoparticle (LNP)‐based mRNA vaccines delivered intramuscularly have been shown to be effective against SARS‐CoV‐2, the virus that caused the coronavirus disease 2019 (COVID‐19) pandemic.^{[ 3} , ^{4 ]} However, alongside the considerable benefits of mRNA therapeutics, severe side effects such as fever, headache, and thrombosis due to the strong immune response induced by cationic lipids have been reported.^{[ 5} , ⁶ , ^{7 ]} Moreover, LNPs tend to accumulate in the liver and lungs, and expression of therapeutic mRNAs in these organs remains challenging.^{[ 8} , ^{9 ]} While cationic lipids are incorporated as agents to enhance endosomal escape, the efficiency of endosomal escape is approximately 2%, posing a major obstacle to delivery efficiency.^{[ 10} , ^{11 ]} Thus, the development of more efficient and safer non‐LNP‐based mRNA delivery carriers is highly anticipated.

Although coacervate‐based systems and many polymer‐based nanoparticles share the feature of self‐assembly, coacervate‐based systems generally exhibit a higher encapsulation capacity for cargo molecules, owing to the formation of micrometer‐sized phase‐separated domains compared with the nanometer‐sized domains formed by self‐assembled polymers.^{[ 12 ]} In addition, coacervates are typically formed through electrostatic and hydrophobic interactions. Large numbers of cargo molecules with diverse physicochemical properties can thus be easily accumulated through such interactions. Notably, coacervates can establish much larger contact domains with cell surfaces compared with nanoparticles, which may enable unique modes of intracellular delivery, rather than the canonical endocytic uptake pathway employed by nanoparticles.

Coacervate‐based delivery systems, wherein mRNAs spontaneously form droplet‐like particles by simple mixing with cationic polymers or polypeptides, may offer various design possibilities such as easy and high mRNA loading and controlled release in target organs or cells. Biocompatibility and prolonged therapeutic retention are also expected.^{[ 13} , ¹⁴ , ¹⁵ , ¹⁶ , ^{17 ]} Recent reports using micrometer‐sized particles comprising self‐assembling histidine‐rich peptides support the promise of a coacervate‐based system for mRNA delivery into cells.^{[ 18} , ¹⁹ , ²⁰ , ^{21 ]} However, presumably due to the intrinsic features of coacervates, where electrostatic interactions play a role in complex formation, coacervates are generally unstable under physiological salt conditions (e.g., ∼150 mM NaCl). Few reports have clearly demonstrated the feasibility of intracellular mRNA delivery in serum‐containing media, which is an important requisite for future in vivo animal studies.

Our research group developed coacervate‐based delivery systems that allow the efficient condensation and delivery of immunoglobulin G (IgG) antibodies into cells with remarkable efficiency.^{[ 22} , ²³ , ^{24 ]} These systems consist of multimers of L17E membrane‐permeabilizing peptides^{[ 25 ]} and IgGs with negative charges. The attachment of the coacervate achieved facile cytosolic IgG infusion throughout the cells within 1 min. The modes of intracellular IgG delivery using our coacervate system were also different from those of other systems. While the coacervate‐mediated delivery systems reported by other researchers generally employ endocytic uptake and endosomal escape for intracellular delivery, the cytosolic translocation of IgG most likely occurs either directly through the plasma membrane or at a very early stage of endocytosis, accompanied by dynamic structural alterations of the cytoskeletal protein F‐actin and the plasma membrane.^{[ 22} , ²³ , ^{24 ]} This delivery can be achieved at physiological salt concentrations in the presence of serum. If this efficient mode of cytosolic infusion is applicable to cytosolic mRNA delivery, it should introduce novel concepts for mRNA delivery with therapeutic potential. However, although both are negatively charged, mRNA and these IgG differ greatly in the number and density of charges as well as in their molecular sizes. Given that the balance between electrostatic and hydrophobic interactions plays a critical role in coacervation, it is not straightforward to apply our coacervate‐mediated delivery system to mRNA delivery.

This study aimed to confirm the adaptability of the cationic intracellular delivery peptide FcB(L17E)₃‐based coacervate system for cytosolic mRNA delivery, particularly under physiological salt conditions and in the presence of serum. Considering that the IgG used for coacervate‐mediated delivery carries about 20 negative charges, we first evaluated the applicability of the system for siRNA delivery. Because aggregation tended to occur during the coacervation of mRNA (bearing considerably higher negative charges than siRNA) with FcB(L17E)₃, we explored the use of a scaffolding molecule with a negative charge number comparable to that of siRNA to facilitate coacervate formation and incorporate mRNA. We further tested the applicability of this approach to plasmid DNA (pDNA) delivery, which involves far greater charge density and molecular size. Finally, in view of potential future applications in cancer immunotherapy and mRNA‐based vaccination, we examined subcutaneous administration of mRNA‐loaded microcondensates. To highlight the differences in delivery mode and efficiency, we use the term “microcondensate” to describe the FcB(L17E)₃‐based coacervates employed in our system, distinguishing them from other coacervate‐based delivery platforms.

Results and Discussion

Formation of Microcondensates and Cytosolic Infusion of siRNA

In our previously reported microcondensate‐based IgG delivery using the L17E multimer, IgG was modified with approximately 10 molecules of Alexa Fluor 488 (Alexa488).^{[ 23 ]} Considering that Alexa488 contains two sulfonate groups, the total number of negative charges introduced per IgG molecule (∼150 kDa) was estimated to be approximately 20. In contrast, an mRNA encoding a 100‐amino acid protein comprises at least 300 ribonucleotides, excluding untranslated regions. This indicates that the putative mRNA had at least 300 negative charges with a molecular mass of 96 kDa, assuming that the residual mass of the ribonucleotide was ∼320 and that the number or density of negative charges in the mRNA was much higher than that in Alexa488‐modified IgG (IgG‐Alexa488). The balance between the negative and positive charges is critically important for coacervate formation. Excess electrostatic interactions among polyion complexes can lead to the formation of insoluble aggregates,^{[ 26 ]} which are undesirable for mRNA formulation and delivery. In comparison, a typical siRNA molecule (∼14 kDa) carries approximately 40 negative charges, a value comparable to that of IgG‐Alexa488. Therefore, in our first trial, we investigated whether microcondensates could be formed by siRNA in the presence of FcB(L17E)₃, the L17E trimer that we employed for microcondensate formation with IgG‐Alexa488.^{[ 23 ]} As a model siRNA, 5′‐fluorescein‐labeled siRNA targeting β‐actin (si‐β‐actin‐FAM) was employed to investigate the relationship between cellular internalization, localization patterns, and siRNA activity (Figure 1a), a system frequently used to evaluate siRNA function.^{[ 27 ]}

Figure 1.

Formation of microcondensates of siRNA with FcB(L17E)₃ and intracellular delivery of siRNA. a) Schematic diagram of microcondensate formation and intracellular delivery by microcondensate of siRNA and FcB(L17E)₃. b) Confocal laser scanning microscopy (CLSM) images of microcondensates formed between si‐β‐actin‐FAM (siRNA‐FAM) and FcB(L17E)₃ (final concentrations, 5 and 2 µM, respectively) at pH 6.0 and 7.0. The scale bars represent 10 µm. c) Microcondensates formed by mixing FcB(L17E)₃ with varying mixing ratios of si‐β‐actin‐FAM (siRNA‐FAM) in MBS (25 mM MES containing 150 mM NaCl, pH 6.0) with the final concentration of FcB(L17E)₃ of 1 µM and si‐β‐actin‐FAM as indicated. The scale bars represent 10 µm. d) CLSM images of HeLa cells treated with microcondensates formed by si‐β‐actin‐FAM (siRNA‐FAM) and FcB(L17E)₃ (final concentrations were given in the figure). The scale bars represent 50 µm. e) Time‐lapse imaging of Lifeact‐mCherry‐expressing HeLa cells treated with microcondensates formed by si‐β‐actin‐FAM and FcB(L17E)₃ (final concentrations: 1 and 2.5 µM, respectively). The scale bar represents 20 µm. f) Knockdown of β‐actin within cells treated with microcondensates formed with nonlabeled si‐β‐actin or negative control siRNA (si‐nega.) in the presence and absence of FcB(L17E)₃. Error bars represent the mean ± the standard error of mean (SEM) (n = 3). Transfection using Lipofectamine RNAiMax (LF‐RM) was conducted following the protocol by the distributor. Statistical analysis was conducted using a one‐way analysis of variance (ANOVA), followed by Dunnett's test; *p < 0.05 and ****p < 0.0001 versus si‐nega. with LF‐RM.

To assess microcondensate formation under physiological salt concentrations and explore potential pH effects si‐β‐actin‐FAM and FcB(L17E)₃ were incubated in 25 mM 2‐(N‐morpholino)ethanesulfonic acid (MES), with or without 150 mM NaCl, at pH 6.0 or 7.0 at 23 °C for 30 min (Figure 1b). Confocal microscopy revealed that in the absence of NaCl, si‐β‐actin‐FAM and FcB(L17E)₃ formed aggregates, lacking the uniform, droplet‐like morphology and exhibiting low mobility (>10 µm) at both pH values, likely due to the strong electrostatic interactions between si‐β‐actin‐FAM and FcB(L17E)₃ (Figure 1b, 150 mM NaCl(−)).^{[ 28 ]} However, microcondensates of several micrometers in diameter were clearly observed in 25 mM MES containing 150 mM NaCl (pH 6.0) (hereafter referred to as MES‐buffered saline (MBS)) (Figure 1b, 150 mM NaCl(+), pH 6.0). Aggregates were predominantly formed at pH 7.0 even in the presence of 150 mM NaCl (Figure 1b, 150 mM NaCl(+), pH 7.0). These results suggested the presence of an optimum range of electrostatic interactions for microcondensate formation. Excessive interaction in the absence of 150 mM NaCl led to the aggregate formation of si‐β‐actin‐FAM with FcB(L17E)₃. The pK _a values of the RNA phosphate and nucleobases of ∼1 and >5, respectively, did not explain the preference for microcondensate formation at pH 6.0 over 7.0.^{[ 29 ]} The pK _a values of the side chains of ionic amino acids in FcB(L17E)₃ (i.e., glutamate, lysine, and histidine) were reported to be approximately 4, 10.5, and 6, respectively. Therefore, the difference in the protonation state of histidine‐imidazole likely yielded a difference in the morphological alteration of the microcondensate in the presence of 150 mM NaCl at pH 7.0.

A study examining various concentrations of si‐β‐actin‐FAM (2.5, 5, 10, and 20 µM) mixed with a fixed concentration of FcB(L17E)₃ (2 µM) assessed microcondensate formation (Figure 1c). Microcondensates formed efficiently at si‐β‐actin‐FAM concentrations of 2.5 and 5 µM. However, at 10 and 20 µM si‐β‐actin‐FAM, particle size decreased and nonspherical particles formed, presumably because coacervate formation is highly sensitive to the stoichiometric balance between cationic and anionic components,^{[ 30 ]} and excess siRNA‐FAM likely disrupted this balance, leading to fewer droplets. Due to the small particle numbers, the mixture of 20 µM si‐β‐actin‐FAM and 2 µM FcB(L17E)₃ was excluded from further cellular experiments.

To evaluate cytosolic delivery of siRNA into living cells, microcondensates formed with FcB(L17E)₃ were prepared using the mixing ratio of si‐β‐actin‐FAM and FcB(L17E)₃ shown in Figure 1c. These microcondensates were then applied to HeLa cells under physiological salt conditions in the presence of serum. Confocal microscopy revealed efficient intracellular distribution of siRNA, supporting the effectiveness of the microcondensate‐based delivery approach (Figure 1d). After incubating si‐β‐actin‐FAM with FcB(L17E)₃ in MBS (pH 6.0) for 30 min, the mixture was diluted fourfold with a serum‐containing medium to yield a final FcB(L17E)₃ concentration of 1 µM and final pH of 7.6 (Figure S1). When applied to HeLa cells, strong cytosolic distribution of si‐β‐actin‐FAM was observed within 1 h using microcondensates prepared from 2.5 µM si‐β‐actin‐FAM and 1 µM FcB(L17E)₃. A reduced signal was observed with microcondensates prepared from 1.25 and 5 µM si‐β‐actin‐FAM and 1 µM FcB(L17E)₃. These results align with the microcondensate‐forming efficiency shown in Figure 1c, highlighting the importance of microcondensate formation in achieving efficient siRNA delivery into cells. Considering that FcB(L17E)₃ has 24 positive charges per molecule (i.e., FcB(L17E)₃ should have 27 positive charges derived from N‐termini amino functions and the side chains of lysine, arginine, and histidine residues, from which negative charges derived from glutamate was subtracted), the 2 µM concentration of FcB(L17E)₃ corresponds to 48 µM of positive charge. Microcondensates were formed using 5 µM siRNA (21 nucleotide pairs; 200 µM as phosphoester, considering the absence of 5′‐ and 3′‐end phosphates) and 2 µM FcB(L17E)₃. Given that FcB(L17E)₃ has 24 positive charges per molecule, the negative/positive charge (N/P) ratio was calculated as 200:48 ≈ 4.2:1. In contrast, direct mixing of si‐β‐actin‐FAM and FcB(L17E)₃ in a medium containing both MBS (pH 6.0) and α‐MEM(+) without preincubation failed to produce visible microcondensates or detectable intracellular fluorescence (Figure S2).

Time‐lapse imaging revealed that si‐β‐actin‐FAM delivery began at the cell periphery and completed cytosolic distribution within 1 min after onset (Figure 1e, FAM, 25–26 min). The cytoskeletal protein F‐actin, stained with Lifeact‐mCherry,^{[ 26 ]} showed significant structural alteration immediately before siRNA infusion, beginning after microcondensate attachment to the plasma membrane (Figure 1e, mCherry, 24–25 min). Similar membrane dynamics were observed previously with FcB(L17E)₃‐mediated delivery of IgG‐ALexa488 or negatively charged IgG.^{[ 22} , ^{28 ]}

To assess gene silencing, microcondensates were prepared from unlabeled si‐β‐actin and FcB(L17E)₃, and HeLa cells were treated for 24 h. Then, mRNA was extracted, and β‐actin gene expression was analyzed using real‐time quantitative PCR. A decrease in β‐actin mRNA expression of approximately 40% was observed in cells treated with microcondensates formed by 2.5 µM si‐β‐actin and 1 µM FcB(L17E)₃, compared to those treated with microcondensates formed by 2.5 µM negative control siRNA and 1 µM FcB(L17E)₃ (Figure 1f). These results were obtained in a serum‐containing medium whose properties have not been clearly demonstrated in previous studies of mRNA delivery.^{[ 20 ]} A similar extent of knockdown was obtained using si‐β‐actin and Lipofectamine RNAiMax (LF‐RM; a commercially available cationic lipid for siRNA transfection); however, the microcondensate needed a larger amount of si‐β‐actin compared to that of LF‐RM (Figure 1f, 0.01 µM si‐β‐actin/LF‐RN). LF‐RM is known as a highly efficient transfection agent for siRNA and thus was used here as a benchmarking control. No significant suppression of gene expression was observed when si‐β‐actin and FcB(L17E)₃ were directly mixed in MBS (pH 6.0) and α‐MEM(+) without preincubation (Figure 1f, 2.5 µM si‐β‐actin/1 µM FcB(L17E)₃, Mix).

Properties of Microcondensates Formed from siRNA and FcB(L17E)₃

Electrostatic interaction is a critical factor in the formation of microcondensates. Therefore, microcondensates are generally unstable under high salt conditions. The stability of microcondensates generated from si‐β‐actin‐FAM and FcB(L17E)₃ was thus evaluated in terms of the turbidity of the mixture analyzed by the absorbance at 650 nm (expressing turbidity, which reflects particle number). An increase in the NaCl concentration led to a significant decrease in the absorbance of the mixture at 650 nm (Figure 2a), suggesting that the microcondensates were formed by liquid–liquid phase separation due to electrostatic interactions. However, as absorbance decreased by only ∼40% even at 600 mM NaCl, this suggests that hydrophobic interactions also contribute to microcondensate stability. The fluidity of si‐β‐actin‐FAM inside the microcondensate was assessed using fluorescence recovery after photobleaching (FRAP). No substantial fluorescence recovery was observed even 5 min after photobleaching (Figure 2b,c), indicating low mobility of si‐β‐actin–FAM inside the microcondensates. A similar tendency was observed in our previous studies of microcondensates formed with negatively charged IgGs and FcB(L17E)₃, suggesting that nucleic acid‐containing microcondensates also possess a gel‐like property.^{[ 22} , ^{23 ]} To further evaluate siRNA retention, microcondensates composed of si‐β‐actin–FAM and FcB(L17E)₃ were diluted in MES‐buffered saline(pH) (MBS, 25 mM MES containing 150 mM NaCl), and release of si‐β‐actin–FAM was monitored by confocal microscopy at 30 min, 1 h, and 3 h post‐dilution (Figure 2d). The fluorescence intensity within the microcondensates showed no substantial decrease for up to 3 h, indicating that siRNA remains associated with the microcondensates during prolonged incubation under physiological salt conditions.

Figure 2.

Salt stability of the microcondensates and siRNA mobility. a) Measurement of absorbance at 650 nm to assess microcondensate formation by si‐β‐actin‐FAM and FcB(L17E)₃ under increasing salt concentrations. Error bars represent mean ± SEM (n = 3). b) Representative time‐lapse confocal microscopy images of microcondensates formed by si‐β‐actin‐FAM and FcB(L17E)₃ before and after fluorescence bleaching. Scale bar, 2 µm. Time zero represents just after laser irradiation. c) Decrease in fluorescent intensity of microcondensates containing si‐β‐actin‐FAM and FcB(L17E)₃ after laser irradiation (arrow). d) Confocal microscopy images of microcondensates containing siRNA‐FAM after the addition of HBS, observed at the indicated time points. Scale bars represent 10 µm.

Intracellular Delivery of mRNA Using Microcondensates Formed with FcB(L17E)₃ in the Presence of Single‐Strand DNA (ssDNA)

Currently, mRNA delivery is a major focus of therapeutic research; however, the large size and high negative charge density of mRNA often lead to strong multivalent interactions with cationic peptides, resulting in aggregate formation rather than coacervates. We hypothesized that introducing a short, flexible single‐stranded DNA (ssDNA) scaffold for mRNA could modulate peptide–mRNA interactions, thereby promoting microcondensate formation with FcB(L17E)₃ and enabling efficient cytosolic delivery across different mRNA lengths (Figure 3a). To test this hypothesis, we investigated whether the microcondensate‐mediated approach could be directly applied to mRNA delivery. Specifically, we examined the formation of microcondensates using FcB(L17E)₃ and mRNAs of varying lengths as well as subsequent intracellular delivery of these mRNAs. Three types of mRNA, namely, 311‐nucleotide mRNA encoding HiBiT (HiBiT‐mRNA), 992‐nucleotide mRNA encoding enhanced green fluorescent protein (EGFP‐mRNA), and 1925‐nucleotide mRNA encoding luciferase (Luc‐mRNA), were prepared through in vitro transcription.

Figure 3.

Effects of ssDNA on microcondensate formation of HiBiT‐mRNA with FcB(L17E)₃ and peptide expression through the mRNA delivery into cells. a) Possible delivery of mRNA into cells using microcondensates formed with FcB(L17E)₃ in the presence of ssDNA. b) Schematic illustration of peptide expression assay using mRNA encoding HiBiT peptide (HiBiT‐mRNA). After transfection with microcondensates containing HiBiT‐mRNA, cells were lysed, and the resulting lysate containing expressed HiBiT was combined with extracellularly prepared LgBiT to reconstitute NanoLuc luciferase. After the addition of the substrate, luciferase activity was analyzed. c) Formation of microcondensates by ssDNA with FcB(L17E)₃. d) Confocal microscopy images of microcondensate formed by rhodamine‐labeled HiBiT‐mRNA with FcB(L17E)₃ in the presence of ssDNA. Final concentrations of FcB(L17E)₃ and ssDNA were 2 and 10 µM, respectively. Scale bars represent 10 µm. e) Quantification of HiBiT expression in cells treated with microcondensates formed by HiBiT‐mRNA and FcB(L17E)₃ in the presence of ssDNA by chemiluminescence. Final concentrations of FcB(L17E)₃ and ssDNA were 1 and 5 µM, respectively. Transfection using Lipofectamine MessengerMax (LF‐MM) was conducted following the protocol by the distributor. Error bars represent mean ± SEM (n = 3).

As shown in Figure 1c, microcondensates were formed from siRNA with FcB(L17E)₃ at the final concentrations of 5 and 2 µM, respectively. For comparison, HiBiT‐mRNA was mixed with FcB(L17E)₃ at concentrations of 0.1 and 2 µM, respectively, to make the total negative charge derived from the mRNA mixture become comparable to that of si‐β‐actin. However, this combination produced only a small number of particle‐like structures, far fewer than observed for siRNA (Figure S3, mRNA = 0.1 µM). The HiBiT peptide, part of the split nanoluciferase system, allows indirect evaluation of mRNA delivery by measuring luciferase activity following reconstitution with LgBiT (Figure 3b). However, the addition of the mixture to the cell culture resulted in only a marginal increase in luciferase activity (Figure 3e, mRNA = 50 nM, FcB(L17E)₃ = 1 µM). An increase in the concentration of HiBiT‐mRNA in mixtures with FcB(L17E)₃ predominantly led to the formation of aggregate structures (Figure S3, mRNA = 0.5 and 1 µM, FcB(L17E)₃ = 1 µM). HiBiT‐mRNA has a longer chain length (311 nucleotides; 311‐nt) compared to siRNA (21 base pairs). This provides a much greater number of electrostatic and hydrophobic interaction sites with FcB(L17E)₃, resulting in more extensive networks of nucleic acids and peptides that are prone to aggregation. Because aggregation tends to reduce delivery efficiency,^{[ 24 ]} we sought to mitigate the interaction between mRNA and FcB(L17E)₃.

Given our successful formation of microcondensates composed of siRNA and FcB(L17E)₃, we explored the possibility of using ssDNA instead of siRNA to form microcondensates with FcB(L17E)₃. We hypothesized that the presence of a small amount of mRNA together with ssDNA might still permit microcondensate formation with FcB(L17E)₃, thereby enabling mRNA delivery into cells (Figure 3a). Although ssDNA may also be delivered into cells, it can be readily degraded without causing notable side effects. A 17‐mer ATGC‐repeat ssDNA, which was designed to balance base composition and minimize unwanted hybridization with cellular nucleic acids, was therefore employed (Figure S5A). Ultimately, spherical microcondensates were readily formed between ssDNA and FcB(L17E)₃ (Figure 3c).

The capability of HiBiT‐mRNA and ssDNA to form condensates in the presence of FcB(L17E)₃ was confirmed via CLSM observation 30 min after the incubation of HiBiT‐mRNA, ssDNA, and FcB(L17E)₃ in MBS (pH 6.0; Figure 3d, DIC). Final concentrations of ssDNA and FcB(L17E)₃ were set at 10 and 2 µM, respectively, and varying concentrations of rhodamine‐labeled HiBiT‐mRNA (final concentrations of 0.01, 0.05, and 0.1 µM) were added to the mixture. Entrapment of HiBiT‐mRNA in the microcondensates was also confirmed by the overlapping of rhodamine signals with spherical structures in the DIC channel (Figure 3d, merge). At lower mRNA concentrations (0.01 µM), rhodamine fluorescence was observed only in part of each microcondensate, whereas at higher concentrations (0.1 µM), the fluorescence signal was distributed throughout the entire microcondensate. These observations suggest that the proportion of mRNA incorporated into the microcondensate increases with mRNA concentration. The resulting microcondensates delivered mRNA into the cells (Figure 3e). Microcondensates were similarly formed with HiBiT‐mRNA and FcB(L17E)₃ in the presence of ssDNA using HiBiT‐mRNA without rhodamine labeling. The cells were incubated in serum‐containing medium for 1 h. After washing, the cells were further incubated for 23 h, and HiBiT expression was determined by reconstitution with LgBiT. The final concentrations of ssDNA and FcB(L17E)₃ were set at 5 and 1 µM, respectively. HiBiT expression increased in a concentration‐dependent manner as the amount of HiBiT‐mRNA added to the cells increased. Almost 25% of HiBiT expression was attained using 50 nM HiBiT‐mRNA, compared to that of the use of Lipofectamine MessengerMax (commercially available cationic lipid for mRNA transfection, LF‐MM). The N/P ratio was calculated to be 193:48 (≈4.0:1) when 10 µM 17‐nt ssDNA (160 µM as phosphoester) and 100 nM HiBiT‐mRNA (33 µM as phosphoester) were combined with 2 µM FcB(L17E)₃.

Interestingly, signals from rhodamine‐labeled HiBiT‐mRNA were observed in most microcondensates formed from 10 µM ssDNA and 2 µM FcB(L17E)₃ in the presence of mRNA, regardless of the mRNA concentration used (0.01–0.1 µM) (Figure 3d). However, luciferase activity resulting from microcondensate treatment of cells showed a marked dependence on mRNA concentration (Figure 3e). Notably, at 0.01 µM rhodamine‐labeled HiBiT‐mRNA, the rhodamine signals were mainly confined to a limited region within each microcondensate, whereas at 0.05 and 0.1 µM, the signals appeared distributed throughout the microcondensates. It also appeared that the size of the microcondensates decreased with increasing concentrations of rhodamine‐labeled HiBiT‐mRNA in the mixture (Figure 3c,d). These observations suggest that the presence of highly concentrated mRNA within microcondensates without aggregation is important for microcondensate‐mediated mRNA delivery into cells. This may be because only mRNA localized within a certain fraction of microcondensates is actually delivered into cells upon attachment of the microcondensates to the plasma membranes.

The feasibility of the microcondensate‐based mRNA delivery system in the presence of ssDNA was confirmed using EGFP‐mRNA. The formation of microcondensates was confirmed from EGFP‐mRNA at the final concentrations of 0.01 and 0.1 µM in the presence of FcB(L17E)₃ and ssDNA at the final concentrations of 2 and 10 µM, respectively (Figure 4a), mRNA = 0.01 and 0.1 µM). Even in the presence of ssDNA, increasing the concentration of EGFP‐mRNA to the final concentration of 0.5 µM or the interaction with FcB(L17E)₃ resulted in substantial increase in aggregate formation (Figure 4a, mRNA = 0.5 µM). Aggregate formation was predominant in the absence of ssDNA, despite the use of EGFP‐mRNA at the final concentrations of 0.01 or 0.1 µM (Figure S4A). Treating the cells with microcondensates corresponding to those in (Figure 4a) in the middle resulted in significant EGFP expression at 23 h after incubation (Figure 4b, middle). Flow cytometry quantification showed EGFP expression in approximately 23% of the cells (Figure 4c, second bar from the right and S4B, left), albeit at less than half the efficiency achieved by LF‐MM (Figure 4b,c, right). The N/P ratio was calculated to be 169.9:48 (≈3.5:1) when 10 µM 17‐nt ssDNA (160 µM as phosphoester) and 10 nM EGFP‐mRNA (9.9 µM as phosphoester) were combined with 2 µM FcB(L17E)₃. No significant EGFP expression was detected in the absence of ssDNA due to aggregate formation (Figure 4b(left),c(central bar)).

Figure 4.

Effects of ssDNA on microcondensate formation of EGFP‐mRNA with FcB(L17E)₃ and protein expression through mRNA delivery into cells. a) DIC images of microcondensate formed by EGFP‐mRNA with FcB(L17E)₃ in the presence of ssDNA in MBS (pH 6.0). Final concentrations of FcB(L17E)₃ and ssDNA were set as 2 and 10 µM, respectively. Scale bars represent 10 µm. b) Confocal microscopy images of HeLa cells treated with microcondensates. Scale bars represent 50 µm. c) Flow cytometry analysis quantifying the percentage of EGFP‐expressing cells after microcondensate treatment. Transfection using LF‐MM was conducted following the protocol by the distributor. Error bars represent mean ± SEM (n = 3). d) Confocal microscopy images of microcondensates containing ssDNA, EGFP‐mRNA, and FcB(L17E)₃ after the addition of serum‐containing medium (final serum concentration: 7.5 (v/v)%), observed at the indicated time points (white arrowheads indicate microcondensates). Scale bars represent 10 µm. e) Number of microcondensate particles formed by ssDNA, EGFP‐mRNA, and FcB(L17E)₃ and incubated in serum‐containing medium for the indicated time, calculated from bright‐field images. Data represent the average of four fields of view. f) Confocal microscopy images of HeLa cells treated with microcondensates incubated in serum‐containing medium for the indicated time. Scale bars represent 50 µm.

Given that nucleases and proteases in serum may degrade microcondensate structures or the encapsulated mRNAs, we examined the stability of EGFP‐mRNA‐containing microcondensates in serum‐containing medium. EGFP‐mRNA‐loaded microcondensates were prepared and incubated in serum‐containing medium for 30 min, 1 h, or 3 h, followed by evaluation of particle morphology and number. Under all conditions, the microcondensates maintained a spherical shape with comparable particle counts (Figure 4d,e), demonstrating substantial stability in the presence of serum. When these serum‐preincubated microcondensates (1 h or 3 h) were applied to cells and EGFP expression was evaluated by CLSM 23 h later, weaker but still significant EGFP signals were observed for the 1 h condition, whereas considerably diminished signals were observed after 3 h of pretreatment, compared with cells treated immediately after preparation (Figure 4f). These findings suggest that mRNA/ssDNA/FcB(L17E)₃ microcondensates retain droplet‐like structures under serum‐containing conditions but gradually lose their mRNA delivery activity within a few hours, presumably due to replacement by serum proteins, such as albumin, or degradation by RNases present in serum. Nonetheless, the initially high delivery activity of the microcondensates may compensate for this time‐dependent decline.

As described in the Introduction, a considerable number of coacervate‐based delivery systems for mRNA and siRNA have been developed to date, including systems utilizing the self‐assembly of histidine‐rich peptides.^{[ 18} , ¹⁹ , ^{20 ]} However, only a few reports have clearly demonstrated their in vivo applicability for the delivery of micoRNA and antibodies,^{[ 31} , ^{32 ]} presumably due to the intrinsic instability of coacervates under physiological salt conditions, as electrostatic interactions play a crucial role in maintaining coacervate structures. Additionally, some systems claim the capability of incorporating “universal” types of macromolecules into coacervates for delivery. This also suggests the possible incorporation of various serum‐derived molecules into these coacervates, potentially leading to cargo replacement or co‐delivery of contaminants into cells. In contrast, FcB(L17E)₃ preferentially forms metastable microcondensates with macromolecules rich in negative charges, thereby enabling more specific delivery. These features should be beneficial for in vivo applications and enhance the impact of our microcondensate‐based delivery system.

Furthermore, to investigate the effect of the sugar backbone, we evaluated microcondensate formation using ssDNA and ssRNA of equivalent sequences (Figure S5A). CLSM analysis revealed that ssRNA formed smaller microcondensates than ssDNA with FcB(L17E)₃ under identical condition (Figure S5B, 10 nM EGFP‐mRNA (−)). This tendency was more pronounced when ssRNA and EGFP‐mRNA were mixed with FcB(L17E)₃; significantly smaller particles were formed compared to those formed using ssDNA, EGFP‐mRNA, and FcB(L17E)₃ (Figure S5B, 10 nM EGFP‐mRNA (+)). These results suggest that the presence of 2′‐OH group of ribose in RNA influences the microcondensate formation, possibly due to the difference in resulting hydrogen‐bonding patterns or conformational structures.

To examine whether mRNA can be delivered into cells via microcondensates with ssRNA, microcondensates composed of ssRNA, EGFP‐mRNA, and FcB(L17E)₃ were prepared and applied to HeLa cells, followed by observation using confocal microscopy. Although a small number of cells exhibited faint fluorescence, the expression levels were found to be extremely low (Figure S5C). These observations indicate that differences in sugar backbone chemistry can significantly impact coacervate morphology.

Microcondensate‐mediated mRNA delivery was also applicable to the 1925‐nucleotide luciferase mRNA (Luc‐mRNA) in the presence of ssDNA. A mixture of Luc‐mRNA and FcB(L17E)₃ was incubated in the presence of ssDNA in MBS (pH 6.0) for 30 min before CLSM observation. The final concentrations of FcB(L17E)₃ and ssDNA were again set at 2 and 10 µM, respectively, and Luc‐mRNA concentration varied from 2.5 to 500 nM. Microcondensates with spherical structures were observed at 2.5 and 5 nM Luc‐mRNA, and aggregates were predominantly formed at higher Luc‐mRNA concentrations of 50 and 500 nM (Figure 5a). Owing to the increase in chain length, aggregates were more easily formed at lower mRNA concentrations compared to that of concentrations of HiBiT‐mRNA and EGFP‐mRNA. In contrast, no significant aggregate or microcondensate formation was observed with 2.5 or 5 nM Luc‐mRNA in the absence of ssDNA (Figure S6). Additionally, it is known that mRNAs with higher GC content possess more stable secondary structures.^{[ 33 ]} HiBiT‐mRNA, EGFP‐mRNA, and Luc‐mRNA exhibit GC contents of 39%, 56%, and 62%, respectively (Table S1). Since the tendency for aggregate formation of the mRNAs used in this study correlates with both chain length and GC content, further studies are warranted to critically evaluate the effects of nucleic acid composition and RNA secondary structure on microcondensate formation.

Figure 5.

Effects of ssDNA on microcondensate formation of Luc‐mRNA and FcB(L17E)₃ and protein expression through Luc‐mRNA delivery into cells. a) DIC images of microcondensates formed by Luc‐mRNA and FcB(L17E)₃ in the presence of ssDNA. Scale bars represent 10 µm. b) Quantification of luminescence in cells treated with microcondensates formed by Luc‐mRNA and FcB(L17E)₃ in the presence of ssDNA. Transfection using LF‐MM was conducted following the protocol by the distributor. Error bars represent mean ± SEM (n = 3). c) Measurement of absorbance at 650 nm to assess microcondensate formation by ssDNA and FcB(L17E)₃ or mixtures of Luc‐mRNA, ssDNA, and FcB(L17E)₃ under increasing salt concentrations. Final concentrations of ssDNA and FcB(L17E)₃: 15 µM and 3 µM, respectively. Error bars represent mean ± SEM (n = 3).

Treating cells with microcondensates containing Luc‐mRNA resulted in luciferase expression, and luciferase activity was approximately 30% of that attained using Lipofectamine MessengerMax (Figure 5b). The N/P ratio was calculated to be 169.6:48 (≈3.5:1) when 10 µM 17‐nt ssDNA (160 µM as phosphoester) and 5 nM EGFP‐mRNA (9.6 µM as phosphoester) were combined with 2 µM FcB(L17E)₃. Using ssDNA, microcondensates containing mRNAs encoding both peptides (HiBiT) and proteins (EGFP and luciferase) were successfully formed with FcB(L17E)₃ and delivered into cells, leading to the expression of the respective peptide and proteins.

To investigate the salt stability and physical characteristics of the microcondensates, turbidity measurements were conducted at 650 nm by varying the NaCl concentration from 150 to 600 mM in the MES buffer (pH 6.0). Since it was difficult to accurately analyze the absorbance at 650 nm using the concentrations in Figure 5a, microcondensates were prepared with a 1.5‐fold concentration of FcB(L17E)₃, either with ssDNA alone or with Luc‐mRNA and ssDNA, where the ratios of FcB(L17E)₃, ssDNA, and Luc‐mRNA remained the same as in Figure 5a. The final concentrations were 3 µM for FcB(L17E)₃, 15 µM for ssDNA, and 0, 3.75, 7.5, or 75 nM for Luc‐mRNA.

The turbidity of microcondensates formed by ssDNA and FcB(L17E)₃ in the absence of Luc‐mRNA at 150 mM NaCl was approximately 0.08, which decreased markedly to 0.02 at 300 mM NaCl (Figure 5c, ssDNA). A similar tendency was observed when the turbidity was analyzed in the presence of Luc‐mRNA at the concentrations corresponding to 2.5 or 5 nM mRNA in Figure 5a (Figure 5c, ssDNA + 3.75 or 7.5 nM mRNA). Under these conditions, microcondensates with spherical structures were predominantly formed. However, at the Luc‐mRNA concentration corresponding to 50 nM mRNA in Figure 5a, where aggregate formation became prominent (Figure 5c, ssDNA + 75 nM mRNA), no marked decrease in turbidity was observed at 300 mM NaCl concentrations. These results suggest that electrostatic interactions are a dominant factor in organizing microcondensate structures at the mRNA concentrations of 2.5 and 5 nM in Figure 5a. An increase in the Luc‐mRNA concentration to 75 nM (corresponding to 50 nM mRNA in Figure 5a) leads to increased interaction with FcB(L17E)₃, and factors other than electrostatic interactions (e.g., hydrophobic interactions) tend to form aggregates—excess mRNA disrupts the balance of electrostatic/hydrophobic interactions necessary for microcondensate formation with spherical structures.

The contribution of hydrophobic interactions to the stability of nucleic acid‐loaded microcondensates was evaluated by turbidity measurements and microscopic observations following treatment with 1,6‐hexanediol (1,6HD) (Figure S7A,B). Since 1,6HD is known to disrupt weak hydrophobic interactions involved in liquid–liquid phase separation (LLPS),^{[ 34 ]} it promotes the dissolution of coacervates stabilized by such interactions. Treatment of pre‐formed ssDNA/FcB(L17E)₃ microcondensates with 5% or 10% 1,6HD caused a slight decrease in absorbance and a moderate reduction in the number of visible particles in both bright‐field and fluorescence images. In contrast, treatment with 20% 1,6HD caused a pronounced decrease in absorbance at 650 nm, indicating reduced turbidity and reflecting a decrease in particle number. Microscopy further revealed an almost complete loss of particles. These results indicate that ssDNA/FcB(L17E)₃ microcondensates are stabilized not only by electrostatic but also by hydrophobic interactions, consistent with observations for FcB(L17E)₃/IgG‐Alexa488 microcondensates.^{[ 23 ]}

To investigate the mechanism of cellular uptake, we first examined whether nucleic acid‐containing microcondensates are internalized via energy‐dependent cellular uptake pathways or through energy‐independent physicochemical mechanisms across the plasma membrane. Fluorescently labeled ssDNA was mixed with FcB(L17E)₃, with or without EGFP‐mRNA, and observed by confocal microscopy (Figure S8A), yielding microcondensates with morphologies similar to those formed with unlabeled ssDNA (Figure 3c,d). When these fluorescently labeled ssDNA‐containing microcondensates were incubated with cells at 4 °C, a condition known to inhibit energy‐dependent processes, no detectable cytosolic fluorescence was observed, indicating that energy‐dependent cellular uptake is required (Figure S8B). We next investigated the pathway of mRNA delivery via microcondensates by pretreating cells with pharmacological inhibitors targeting specific endocytic pathways: chlorpromazine (CPZ, clathrin‐mediated endocytosis),^{[ 35 ]} nystatin (caveolae‐mediated endocytosis),^{[ 36 ]} methyl‐β‐cyclodextrin (MβCD, lipid raft disruption),^{[ 37 ]} and cytochalasin D (CytD, actin polymerization inhibition).^{[ 38 ]} After 30 min of pretreatment, EGFP‐mRNA‐containing microcondensates were added for 1 h, cells were washed, and EGFP expression was evaluated by confocal microscopy and quantified by flow cytometry 23 h after treatment (Figure 6a,b). CPZ and nystatin showed minimal effects on the proportion of EGFP‐positive cells, whereas CytD and MβCD reduced the proportion by approximately 92% and 80%, respectively, compared with controls. These results suggest that, consistent with findings for negatively charged IgG‐based microcondensates, nucleic acid‐containing microcondensates are primarily internalized using lipid raft‐ and actin‐dependent pathways.^{[ 22 ]} The above results suggest that the cellular uptake mode of microcondensates composed of mRNA/ssDNA/FcB(L17E)₃ shares many similarities with that of negatively charged IgG/FcB(L17E)₃ and is accompanied by dynamic structural alterations of F‐actin and the plasma membrane structures.^{[ 22 ]}

Figure 6.

Mechanism of intracellular delivery of microcondensates formed with ssDNA, mRNA, and FcB(L17E)₃. a) Confocal microscopic analysis of the effect of endocytosis inhibitors on the expression of EGFP of the cells treated by microcondensate composed of ssDNA/EGFP‐mRNA/FcB(L17E)₃. Scale bars represent 10 µm. b) Effect of endocytosis inhibitors on the expression of EGFP of the cells treated with microcondensate composed of ssDNA/EGFP‐mRNA/FcB(L17E)₃. The expression ratio (%) was analyzed by flow cytometry. The results are presented as the mean ± SE (n = 3). Statistical analysis was conducted using a one‐way analysis of variance (ANOVA), followed by Dunnett's test; *p < 0.05, **p < 0.01, and ***p < 0.001 versus control.

To evaluate the cell‐type versatility of mRNA delivery by microcondensates, we assessed the introduction of EGFP‐mRNA or Luc‐mRNA into two additional cancer cell lines: Colon‐26 (colon carcinoma) and A431 (epidermoid carcinoma). In Colon‐26 cells, treatment with EGFP‐mRNA‐containing microcondensates resulted in EGFP expression in 57% of cells, a level comparable to that achieved with LF‐MM (57%) (Figure S9A–C). In A431 cells, the expression level of luciferase encoded by mRNA delivered by microcondensates reached approximately 60% of that obtained with LF‐MM (Figure S9D), confirming the ability of this system to deliver mRNA into multiple cancer cell types beyond HeLa cells.

Intracellular Delivery of Plasmid DNA via Microcondensates

Having successfully demonstrated microcondensate‐based delivery of siRNA and mRNA, we next hypothesized that this strategy could be extended to even larger nucleic acids, such as pDNA. Plasmid DNA is substantially larger than mRNA and siRNA and has a much higher phosphate density, which can lead to overly stable complexes. We reasoned that introducing ssDNA could serve as a flexible binding scaffold, balancing the interaction strength and enabling microcondensate formation with FcB(L17E)₃. To test this hypothesis, we used confocal microscopy to examine whether pDNA encoding EGFP (pEGFP, 4733 base pairs; 9466 phosphoester‐derived negative charges) could form microcondensates with FcB(L17E)₃ in the presence of ssDNA. For comparison, 100 nM HiBiT‐mRNA (331 nucleotides), 10 nM EGFP‐mRNA (992 nucleotides), and 5 nM Luc‐mRNA (1925 nucleotides) successfully formed microcondensates with 2 µM FcB(L17E)₃ together with 10 µM ssDNA, and these were capable of mRNA delivery (Figures 3, 4, 5). The concentrations of phosphoester‐derived negative charges were calculated to be 33, 9.9, and 9.6 µM, respectively. Based on these results, to evaluate the applicability of the microcondensate‐mediated delivery to pDNA, we tested a concentration range of 0.1–10 nM for pEGFP (0.95–95 µM as phosphoester), which covers the concentration ranges in mRNA delivery. pEGFP was mixed with FcB(L17E)₃ and ssDNA to yield final concentrations of 0.1, 1, or 10 nM (pEGFP), 2 µM (FcB(L17E)₃), and 10 µM (ssDNA), respectively. CLSM observations revealed that microcondensates with spherical structures were formed at 0.1 and 1 nM pEGFP, whereas the use of 10 nM pEGFP resulted in predominant aggregate formation (Figure 7a). In the absence of ssDNA, no discernible microcondensates were observed at 0.1 nM pEGFP, and marked aggregate formation was observed at 1 and 10 nM (Figure S10A), highlighting the essential role of ssDNA in controlling the interaction strength of the mixture and microcondensate formation.

Figure 7.

Effects of ssDNA on microcondensate formation of plasmid DNA encoding EGFP (pEGFP) and FcB(L17E)₃ and protein expression through plasmid DNA delivery into cells. a) DIC images of microcondensate formed by mixing pDNA and ssDNA with FcB(L17E)₃ to yield their final concentrations. Scale bars represent 10 µm. b) Confocal microscopy images of HeLa cells treated with microcondensate formed by pDNA, ssDNA, and FcB(L17E)₃. Scale bars represent 50 µm. c) Flow cytometry analysis quantifying the percentage of EGFP‐expressing cells treated with microcondensate. Error bars represent mean ± SEM (n = 3). Transfection using LF‐LTX was conducted following the protocol by the distributor.

CLSM analysis showed that cells treated with the pEGFP‐containing microcondensates (final concentrations of pEGFP, FcB(L17E)₃, ssDNA: 0.5 nM, 1 µM, and 5 µM, respectively) expressed EGFP at levels comparable to those obtained with Lipofectamine LTX (commercially available cationic lipid for plasmid transfection, LF‐LTX)‐mediated transfection (Figure 7b). Flow cytometry confirmed that ∼25% of the cell population expressed EGFP following microcondensate treatment, whereas only background‐level expression was observed when pEGFP and FcB(L17E)₃ were mixed without ssDNA (Figure 7c and Figure S10B). Although LF‐LTX treatment led to EGFP expression in approximately 70% of cells, these results demonstrated the feasibility of the FcB(L17E)₃/ssDNA system for intracellular delivery of mRNA and pDNA molecules. The N/P ratio was calculated to be 169.6:48 (≈3.5:1) when 10 µM 17‐nt ssDNA (160 µM as phosphoester) and 1 nM EGFP‐mRNA (9.5 µM as phosphoester) were combined with 2 µM FcB(L17E)₃.

Across siRNA, mRNA, and pDNA, the negative‐to‐positive charge ratio ranged from 3.5:1 to 4.2:1, suggesting that an optimal electrostatic balance is necessary to drive phase separation. Notably, despite the differences in molecular weight and backbone structure between siRNA, ssDNA, mRNA, and pDNA, the optimal charge ratios for microcondensate formation remained almost consistent, suggesting the versatility of this system.

Collectively, these results demonstrate that microcondensates formed from FcB(L17E)₃ in the presence of ssDNA accommodated various nucleic acid types, including mRNA and pDNA, when their charge density is properly tuned. The correlation between droplet morphology, turbidity profiles, and intracellular delivery outcomes supports the concept that microcondensates formed by modular nucleic acid assembly and membrane‐permeabilizing FcB(L17E)₃ hold significant potential as a novel platform for cytosolic delivery of nucleic acids.

In Vivo mRNA Delivery

Finally, we investigated whether mRNA‐bearing microcondensates could deliver mRNA and enable protein expression in vivo. The microcondensates used here had diameters of at least a few micrometers, comparable to the caliber of capillaries (5–10 µm). We envision the most promising applications of microcondensate‐mediated in vivo mRNA delivery to be in cancer immunotherapy and mRNA‐based vaccination via topical or local administration, rather than systemic administration. Therefore, in this study, we focused on subcutaneous administration. The mRNA encoding luciferase used in our previous in vivo evaluation (Luc2‐mRNA)^{[ 39 ]} was employed as a model mRNA. The delivery efficacy was assessed in terms of luciferase activity. Microcondensates were prepared from Luc‐mRNA and FcB(L17E)₃ in the presence of ssDNA at a concentration 2.5‐fold higher than that used for cellular experiments and were administered subcutaneously to mice without dilution. Six hours after administration, luciferase expression was evaluated through in vivo imaging system (IVIS). Luciferase expression was observed at the injection site in mice treated with microcondensates (Figure 8a), with signal intensity exceeding 12‐fold, which is higher than that observed in mice receiving naked mRNA (analyzed by signal intensity from the IVIS image; Figure 8b, microcondensate versus naked mRNA). Importantly, no acute toxicity or mortality was observed in mice under the tested conditions. In mice injected with a mixture of FcB(L17E)₃ and mRNA without ssDNA, luciferase expression was markedly lower than that in mice injected with naked mRNA, presumably because of aggregate formation (Figure 8b, No ssDNA). Although the protein expression level by the use of LNP formulated with ALC0315, an ionizable lipid used in COVID‐19 vaccine Comirnaty,^{[ 40 ]} was nearly 100 times higher than that of microcondensates (Figure S11), this study is the first to demonstrate intracellular mRNA delivery and protein expression in mice using microcondensates or other micrometer‐sized coacervates. The lower in vivo expression efficiency compared to LNPs can be attributed to several key factors related to the stability and behavior of microcondensates in the complex biological environment. First, due to their micrometer‐scale size, microcondensates may be sterically hindered from effectively contacting cells by the extracellular matrix. This compromised their structural integrity and reduced the effective dose reaching target cells. Second, enzymatic degradation by nucleases (e.g., RNases) and proteases present in subcutaneous tissue may have rapidly broken down both the nucleic acid cargo and the peptide component, respectively, prior to cellular uptake and protein expression.

Figure 8.

Protein expression following subcutaneous injection of microcondensates in mice. a) Representative IVIS images following in vivo delivery of the naked mRNA, mixture of FcB(L17E)₃ and mRNA without ssDNA (no ssDNA), and the microcondensate composed of mRNA, ssDNA, and FcB(L17E)₃ (microcondensate) in ICR mice via subcutaneous injection. b) Graph showing radiant efficiency of IVIS‐imaged mice. Error bars represent mean ± SEM (n = 3).

Despite these limitations and the need for further optimization, this study represents the demonstration of in vivo intracellular mRNA delivery and protein expression in mice using a microcondensate‐based approach. Additionally, the unique properties of microcondensates—particularly their ability to induce localized protein expression upon direct injection—offer great potential for applications requiring regional administration, such as mRNA and gene therapy for solid tumors and localized regenerative medicine.

Conclusion

Formation of microcondensates using the cationic peptide FcB(L17E)₃ was first confirmed with siRNA. Intracellular delivery of siRNA and gene knockdown efficiencies comparable to those of commercially available transfection reagents were demonstrated. Analysis of the microcondensate properties using turbidity measurements and FRAP suggested that microcondensates were formed through both electrostatic and hydrophobic interactions and that they exhibited low siRNA fluidity inside, as indicated by the lack of fluorescence recovery. In the case of mRNA, which has longer chains and higher negative charges, aggregation with FcB(L17E)₃ tended to occur. We were able to avoid this problem by using ssDNA as a scaffold. This approach was successful not only in promoting microcondensate formation with FcB(L17E)₃ and mRNA but also in reducing the mRNA concentration required for protein expression. Using such microcondensates, intracellular delivery and protein expression were achieved with HiBiT‐mRNA encoding the HiBiT peptide (311 nucleotides), EGFP‐mRNA (992 nucleotides), and Luc‐mRNA (1925 nucleotides). Importantly, delivery was achieved under physiological salt concentrations and in the presence of serum. Moreover, the successful delivery of plasmid DNA encoding EGFP was also demonstrated, highlighting the applicability of this system to a wide range of nucleic acids. These findings suggest an optimal nucleic acid concentration for microcondensate formation to facilitate intracellular delivery, which ranges from 170 to 200 µM of phosphoester relative to 2 µM of FcB(L17E)₃ (N/P ratio: 3.5–4.2). Studies on the mRNA delivery mechanisms suggested that nucleic acid delivery was attained at the cell surface or at a very early stage of endocytosis, which employs lipid raft‐ and actin‐dependent mechanisms of action accompanied by dynamic structural changes in F‐actin and the plasma membrane. Close similarity in the mechanisms involved in IgG delivery using FcB(L17E)₃‐based microcondensates was also suggested.

While further studies are necessary to clarify the detailed modes of interaction among FcB(L17E)₃, ssDNA, and mRNA in microcondensate formation, these three components together facilitate the formation of micrometer‐sized coacervates containing mRNA, enabling protein expression via mRNA delivery in vivo in mice. The subcutaneous administration of microcondensates containing Luc‐mRNA to mice resulted in >10‐fold higher protein expression than naked mRNA. In the absence of ssDNA, the mixed sample of mRNA and FcB(L17E)₃ exhibited even lower expression compared with naked mRNA, underscoring the importance of ssDNA as a scaffold for forming microcondensates with FcB(L17E)₃ in intracellular mRNA delivery.

In summary, our findings establish a lipid‐free, microcondensate‐based delivery strategy that combines high efficiency with a unique membrane‐translocating mechanism distinct from conventional endocytosis‐dependent coacervates. This system offers a promising new platform for safe and effective nucleic acid delivery, with potential applications in mRNA vaccines and cancer immunotherapy.

Conflict of Interests

Y.K., A.K., and S.F. are inventors on a patent application related to the technology described in this manuscript. All other authors declare no conflict of interest.

Supporting information

Supplementary Information

Kawaguchi Y., Kikkawa A., Kimura S., Abe H., Futaki S., Angew. Chem. Int. Ed. 2026, 65, e12139. 10.1002/anie.202512139

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.