Cytosolic protein delivery using pH-responsive, charge-reversible lipid nanoparticles

Although proteins have attractive features as biopharmaceuticals, the difficulty in delivering them into the cell interior limits their applicability. Lipid nanoparticles (LNPs) are a promising class of delivery vehicles. When designing a protein delivery system based on LNPs, the major challenges include: (i) formulation of LNPs with defined particle sizes and dispersity, (ii) efficient encapsulation of cargo proteins into LNPs, and (iii) effective cellular uptake and endosomal release into the cytosol. Dioleoylglycerophosphate-diethylenediamine (DOP-DEDA) is a pH-responsive, charge-reversible lipid. The aim of this study was to evaluate the applicability of DOP-DEDA-based LNPs for intracellular protein delivery. Considering the importance of electrostatic interactions in protein encapsulation into LNPs, a negatively charged green fluorescent protein (GFP) analog was successfully encapsulated into DOP-DEDA-based LNPs to yield diameters and polydispersity index of < 200 nm and < 0.2, respectively. Moreover, ~ 80% of the cargo proteins was encapsulated into the LNPs. Cytosolic distribution of fluorescent signals of the protein was observed for up to ~ 90% cells treated with the LNPs, indicating the facilitated endocytic uptake and endosomal escape of the cargo attained using the LNP system.


Results and discussion
Formulation of charge-reversible lipid-based nanoparticles encapsulating a negatively charged green fluorescent protein. Prior to evaluating the utility of LNPs for intracellular delivery, we first examined whether LNP-encapsulating proteins could be formulated using the DOP-DEDA system 23 . Considering that negatively charged siRNA is easily encapsulated into the DOP-DEDA system, the feasibility for encapsulating a negatively charged protein was first examined.
Effective intracellular delivery of NLS-(− 30)GFP attained by DOP-DEDA-based LNPs. The ability of DOP-DEDA-based LNPs for intracellular protein delivery was evaluated using CLSM. HeLa cells were treated with NLS-(− 30)GFP alone (i.e., without encapsulation into LNPs) or NLS-(− 30)GFP-LNPs with protein/lipid mass ratios of 1:10 and 1:50 in serum-containing medium for 6 h (Fig. 3). Cells treated with NLS-(− 30)GFP (2.5 µM) showed very little cytosolic or dot-like punctate signal in the cells, suggesting marginal cellular uptake of the NLS-(− 30)GFP protein (Fig. 3A, left). Concentration of NLS-(− 30)GFP-LNPs (1:10) and (1:50) was set to yield a final NLS-(− 30)GFP concentration of 2.5 µM in the medium (protein concentrations were calculated assuming that all proteins were encapsulated in the LNPs). Marked cytosolic and nuclear NLS-(− 30)GFP signals were observed in the cells treated with NLS-(− 30)GFP-LNPs (1:10) (Fig. 3A, center). Cytosolic release from endosomes is required prior to the translocation of NLS-(− 30)GFP to the nucleus (Fig. 1C). Therefore, nuclear NLS-(− 30)GFP signals are an indication of the cytosolic release of the protein ( Figure S2). Approximately 60% of cells had NLS-(− 30)GFP signals in the nucleus following treatment with NLS-(− 30) GFP-LNPs (1:10) (Fig. 3B). Interestingly, NLS-(− 30)GFP-LNPs (1:50) yielded much less NLS-(− 30)GFP signals in the nucleus, but had punctate signals in the cells, suggesting the importance of the protein/lipid ratio to yield efficient cytosolic protein release (Fig. 3A, right and B). An increase in the DOP-DEDA ratio against NLS-(− 30) GFP does not necessarily lead to cytosolic release of the encapsulated proteins. It should be noted that in the case of nucleic acid delivery (e.g., siRNA and antisense oligonucleotides), the efficiency of endosomal escape is generally estimated to be less than a few percent 36 . Even under treatment conditions yielding the desired cell activity, CLSM analysis of fluorescently labeled nucleic acids often yield punctate cell distribution without spreading throughout the cell, indicating that the majorities of the nucleic acids are trapped in endosomes or form aggregate in cytosol [37][38][39] . The spread cytosolic signals of NLS-(− 30)GFP observed in this study thus support the suitability of the DOP-DEDA-based LNP system for intracellular protein delivery.
The use of cargos bearing highly negative charges is important for its efficient intracellular delivery using DOP-DEDA-based LNPs. This was suggested through the use of LNPs having the same lipid composition Table 1. Physicochemical characterization of NLS-(− 30)GFP-LNP by mixing NLS-(− 30)GFP with lipids at various mass ratios. The lipid mixture was composed of DOP-DEDA, DPPC, and cholesterol at a 45:10:45 molar ratio, and 1 mol% DMG-PEG5k was added. The mass ratios denote NLS-(− 30)GFP-LNP/total lipids (w/w). PdI = polydispersity index. Results are presented as the mean ± standard deviation (SD) of more than three independent experiments.   (Table S3). A marginal level of cytosolic/nuclear localization in NLS-(− 30)GFP was also observed (Fig. 4B).
The importance of DOP-DEDA in the cytosolic/nuclear delivery of NLS-(− 30)GFP was also confirmed through the use of 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE)-based LNPs. DOPE is a lipid frequently employed for LNP formulations as a fusogenic lipid 31 . DOPE and DOP-DEDA share structural similarities through the ethanolamine and diethylenediamine moieties in their head groups, respectively. However, the amino group of DOPE is always positively charged under physiological conditions (pKa of ethanolamine, 9.5) and lacks pH sensitivity. LNPs with the same lipid composition other than the replacement of DOP-DEDA  Results are presented as mean ± standard deviation (SD) (n = 3). n.s., not significant; ***P < 0.001, (one-way analysis of variance [ANOVA] followed by Tukey-Kramer's honestly-significant difference test). www.nature.com/scientificreports/ Endocytosis is an energy-driven cellular event that does not occur at 4 °C 41 . HeLa cells were treated with NLS-(− 30)GFP-LNP at 4 or 37 °C for 6 h. With a marked contrast compared to the treatment at 37 °C, no substantial NLS-(− 30)GFP signals, even those in dot-like, were observed after treatment at 4 °C (Fig. 6A,B). The effect of endocytosis inhibitors, such as pitstop2 (a clathrin-mediated endocytosis inhibitor) 42 , 5-(N-ethyl-N-isopropyl)amiloride (EIPA, an inhibitor of Na + /H + exchanger and membrane ruffling) 43 and wortmannin (a macropinocytosis inhibitor by blocking phosphatidylinocitol-3-kinase (PI3K)) 44 , on the cellular uptake of NLS-(− 30)GFP-LNP was then analyzed (Fig. 6C). The cellular uptake of NLS-(− 30)GFP was evaluated using flow cytometry based on fluorescence intensity. HeLa cells were pre-treated with each endocytosis inhibitor (30 μM Pitstop2, 80 μM EIPA, or 0.5 μM wortmannin using dimethyl sulfoxide as a vehicle) for 30 min at 37 °C and then incubated with NLS-(− 30)GFP-LNP for 1 h at 37 °C in the presence of these inhibitors. After pitstop2 treatment, the cellular uptake of NLS-(− 30)GFP was 15% of that of untreated cells. Marked decreases in NLS-(− 30)GFP uptake were also observed in the presence of EIPA and wortmannin. These results suggest the possible involvement of clathrin-mediated endocytosis and macropinocytosis in the uptake of LNPs. The involvement of these endocytosis pathways has been suggested in the uptake of other LNPs. A detailed study on their similarities and differences would lead to the development of delivery systems with higher efficacy.
As expected from the time-course analysis of the cytosolic release of NLS-(− 30)GFP delivered by DOP-DEDA-based LNPs, endosomal maturation plays a crucial role. Because charge-reversible DOP-DEDA becomes protonated in acidic buffers, the surface of NLS-(− 30)GFP-LNPs should become positively charged along with endosomal maturation 23 . In contrast, the inner leaflet of endosomal membranes could also be negatively charged because of the abundance of negatively charged endosome-specific lipids, including bis(monoacylglycerol)phosphate (BMP) 45,46 . www.nature.com/scientificreports/ To verify that endosomal acidification and the eventual protonation of DOP-DEDA are crucial for cytosolic NLS-(− 30)GFP release in the DOP-DEDA-based LNP system, endosomal acidification was blocked using ammonium chloride (NH 4 Cl) 47 (Fig. 6D,E). HeLa cells were pre-treated with 25 mM NH 4 Cl for 30 min, then the cells were treated with NLS-(− 30)GFP-LNPs in the presence of 25 mM NH 4 Cl for 6 h. The prevention of endosomal acidification after NH 4 Cl treatment was confirmed by the loss of lysotracker signals, which are pH-sensitive dyes and indicators of acidic vesicular compartments (Fig. 6D, lysotracker, NH 4 Cl( +)).
In marked contrast to the cellular images of diffuse cytosolic/nuclear labeling in the absence of NH 4 Cl treatment (Fig. 6D

Conclusions
There are numerous reports on LNP-based nucleic acid delivery. However, few reports have been published on the intracellular delivery of proteins, especially those including precise CLSM analyses of the cellular fates of cargo proteins. In this study, we established an approach in producing LNPs based on the pH-sensitive, charge-reversible lipid DOP-DEDA, attaining efficient protein delivery into cells. The lipids employed for particle formation were composed of DOP-DEDA, DPPC, Chol, and DMG-PEG5k at a 45/10/45/1 molar ratio, where DOP-DEDA was found indispensable to attain efficient cytosolic release of the model cargo protein (NLS-(− 30)GFP). The obtained LNPs had diameters and PdIs in the preferable range for drug delivery (< 200 nm and < 0.2, respectively). DMG-PEG5k played an important role in obtaining these PdIs. Although negative charges were needed in the cargo protein for effective encapsulation into the LNPs, almost 80% of the NLS-(− 30)GFP used in the formulation was incorporated into LNPs. When administered to the cells, successful delivery of NLS-(− 30)GFP-LNPs to the cell interior was observed in ~ 90% of the cells evaluated by CLSM analysis. We do not presume that all NLS-(− 30) GFP has a proper folding structure after being released from LNP into the cytosol. However, the CLSM analysis showed that a significant amount of NLS-(− 30)GFP maintained its active structure, which is important from the perspective of drug delivery. The high ratio of the cargo proteins charged into the LNPs and released into the cytosol suggest the promise of this DOP-DEDA-based LNP system as a vehicle for the intracellular delivery of bioactive proteins.
The possible involvement of endocytosis, including clathrin-mediated endocytosis and macropinocytosis, were suggested as the mechanisms of the cellular uptake of NLS-(− 30)GFP-LNPs. Endosomal acidification also plays a role in the cytosolic release of NLS-(− 30)GFP from endosomes, as was suggested in a time-course study of the cellular distribution of NLS-(− 30)GFP and the prevention of endosome acidification under NH 4 Cl treatment. These studies on uptake mechanisms suggest the validity of our design concept.
One of the future goals of this research is the intracellular delivery of antibodies, including low molecular weight antibodies (such as single chain variable fragments (scFv) and nanobodies). Currently, antibody therapeutics used in clinical practice are limited to targeting molecules outside the cell. If this LNP system could deliver antibodies into the cytosol, it could extend the scope of antibody therapy to molecules inside the cell, potentially leading to the treatment of unmet medical needs, including cancer. It has been reported that nanocarriers bearing diameters of 50-100 nm generally exhibit long blood half-life and preferentially accumulate at solid tumors, without efficiently excreted in the urine or phagocytosed by macrophages 48,49 . The diameter of the LNPs estimated in this study (100-170 nm) was slightly larger than these, but still within the acceptable range, suggesting the potential in vivo applicability of this approach.
This time NLS-(− 30)GFP was employed as a model cargo. The hydrodynamic radius of the green fluorescent protein is reported to be about 2.3 nm 50 , which is consistent with the distance between adjacent lipid monolayers of LNPs estimated from TEM images (about 5 nm) ( Fig. 2A-(iii)), suggesting that NLS-(− 30)GFP may act as an adhesive and form multilayered structures. This also suggests that LNPs may be able to encapsulate larger sized proteins in larger spaced multilayer structures, by further optimizing formulation methods including lipid compositions if necessary. In this study, proteins with negative charges are used for encapsulation into LNPs. Further work is needed to encapsulate cargo proteins that do not have a negative charge. The use of tag sequences with negative charges may be one possible approach.
As mentioned above, there are still challenges to be overcome. Nevertheless, we believe that this study is an important first step towards understanding intracellular protein delivery using LNPs.

Preparation of lipid nanoparticle (LNP
To study the effect of the pH of the citrate buffer on the NLS-(− 30)GFP-LNP formulation, LNPs were prepared using 1 mM citrate at pH 4.5-6.0 similarly as described above (Table S1).
To study the Effect of volume ratios of the aqueous/organic phases on the NLS-(− 30)GFP-LNP formulation, 3 to 10 volumes of aqueous phase (1 mM citrate buffer, pH 5.0) were added to a fixed volume of t-butanol (lipid concentration, 25 mM). The amount of total protein was also fixed to retain a protein/lipid mass ratio of 1:10 (Table S2).
For GFP-LNP formulations were prepared, then, half of the respective LNP samples were adjusted to include 2% SDS and stored at 4 °C, and the other halves were precipitated via ultracentrifugation (100,000 × g, 4 °C, 2 h). The supernatants were removed, and the pellets were dissolved with 2% SDS to match the volume of the non-ultracentrifuged LNP samples. The same volume of LNP samples was applied onto a 10% acrylamide gel and resolved using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Lastly, to detect the NLS-(− 30) GFP encapsulated in LNP formulations, Coomassie Brilliant Blue staining was performed. The encapsulation efficiency was calculated using Eq. (1), where P pellet is the amount of NLS-(− 30)GFP in the pellet after ultracentrifugation and P total is the amount of NLS-(− 30)GFP in the non-ultracentrifuged LNP samples.

Cryogenic transmission electron microscopy (Cryo-TEM).
To observe the morphology of NLS-(− 30)GFP-LNPs, Cryo-TEM images were collected using a JEOL/JEM-2100F(G5). The LNP suspension was concentrated to a final concentration of 20 mg/mL of total lipids. After a small amount (3 -5 μL) of the LNP suspension was placed on a TEM copper grid covered by a porous carbon film, the excess solution on the grid was immediately plunged into liquid propane in a cryofixation apparatus (Reichert KF-80, Leica Microsystems, Wetzlar, Germany) to generate vitreous ice. Then, the ice was transferred onto the specimen stage of a Cryo-TEM operated at an acceleration voltage of 200 kV, and Cryo-TEM images were obtained at liquid helium temperature (4.2 K).