Polyphenolic Nanoparticle Platforms (PARCELs) for In Vitro and In Vivo mRNA Delivery

Abstract

Despite their successful implementation in the COVID-19 vaccines, lipid nanoparticles (LNPs) still face a central limitation in the delivery of mRNA payloads – endosomal trapping. Improving upon this inefficiency could afford improved drug delivery systems, paving the way toward safer and more effective mRNA-based medicines. Here, we present Polyphenolic Nanoparticle Platforms (PARCELs) as effective mRNA delivery systems. In brief, our investigation begins with a computationally guided structural analysis of 1825 discrete polyphenolic structural data points across 73 diverse small molecule polyphenols and 25 molecular parameters. We then generate structurally diverse PARCELs, evaluating their in vitro mechanism and activity, ultimately highlighting the superior endosomal escape properties of PARCELs relative to analogous LNPs. Finally, we examine the in vivo biodistribution, protein expression, and therapeutic efficacy of PARCELs in mice. In undertaking this approach, the goal of this study is to establish PARCELs as viable delivery platforms for safe and effective mRNA delivery.

Graphical Abstract

Maximizing therapeutic efficacy at the lowest possible dose is a fundamental objective of the drug delivery sciences.^1–4 For example, this principle is true for mRNA-based therapies, which leverage LNP technologies as one way to reduce the required dose of mRNA-based drugs.^5–11 Despite their promising role in preventing COVID-19 infections, LNPs suffer from a significant limitation – endosomal trapping.^8,12–19 In essence, this process prevents mRNA from reaching the cytoplasm, thus hindering its translation into the desired therapeutic protein.^20–23 Therefore, overcoming endosomal trapping is crucial for advancing non-viral vector-based mRNA delivery in biomedical applications.^24–30

While many approaches exist to overcome endosomal trapping, one powerful strategy aims to develop novel material platforms whose structural features promote higher levels of endosomal escape.^31–36 Development of these materials requires identifying molecular candidates from diverse and virtually infinite pools of chemical space that may improve endosomal escape. While many classes of these molecules exist, polyphenols (a class of naturally occurring small molecules found in nature) have emerged as a particularly promising group of molecules whose ability to interact with biological systems makes them attractive candidates in the drug delivery sciences.^37–40 For example, polyphenols have been widely employed in addressing various diseases, including cardiovascular disease,^41–43 Alzheimer’s and Parkinson’s disease,^44–46 and cancer, amongst others, highlighting their potential utility as drug delivery agents.^47–49 However, studies that leverage polyphenols to improve the efficacy of mRNA-based drugs, particularly from the standpoint of improving endosomal escape, currently remain underexplored.

Here, we present Polyphenolic Nanoparticle Platforms (PARCELs) as effective mRNA delivery systems (Figure 1). Our study begins with a computationally guided structural analysis of 1825 discrete polyphenolic structural data points to identify critical design parameters to incorporate into PARCEL. Informed by these data, we successfully formulate and characterize the mRNA delivery properties of PARCELs, ultimately evaluating their in vitro performance including intracellular (e.g., FLuc) and secreted (e.g., EPO) protein expression. To further the generalizability of these data, multiple types of mechanistic studies including cellular association, uptake mechanism, intracellular degradation and trafficking studies, and endosomal escape studies are then performed, ultimately highlighting that PARCEL has superior endosomal escape properties to analogous LNPs. Finally, we examine the in vivo biodistribution and protein expression of PARCELs in mice. In undertaking this approach, the goal of this study is to establish PARCEL as a viable platform for mRNA delivery, while more broadly highlighting the utility in synergizing techniques in structural analysis, formulation, and mechanism to afford better therapies.

Figure 1.

Schematic illustration of the overall concept of this manuscript – to develop and understand the functionality of PARCEL.

Given that polyphenols represent a large class of bioactive small molecules, we sought to begin our study by leveraging computationally guided approaches to select a representative class of diverse polyphenols for incorporation into PARCELs. Toward that end, we first generated 1825 discrete polyphenolic structural data points by analyzing 25 physiochemical properties of 73 unique polyphenols using Marvin Sketch (ChemoAxon) and visualized them as a heat map (Figure S1a, b, Table S1).⁵⁰ Building on these analyses, we selected gallic acid (GA), catechin (CAT), epigallocatechin gallate (EGCG), and tannic acid (TA) as representative polyphenols for investigation in PARCEL due to their diverse structural features within the polyphenol family (Figure S1a, b). Principal component analysis (PCA) of our four selected polyphenols was then performed to project the 25-dimensional parameters into 2-dimensional space, highlighting the structural versatility across our selected polyphenols (Figure S1c).⁵⁰

Each representative PARCEL was then formulated using microfluidic approaches by mixing an aqueous phase containing mRNA and an ethanol phase containing a clinically relevant ionizable lipid (either Moderna’s SM-102 or Pfizer/BioNTech’s ALC-0315),^51–53 a phospholipid (DOPE),⁵⁴ cholesterol,^55,56 a PEG lipid (C14-PEG-2000),⁵⁷ and a polyphenol (GA, CAT, EGCG or TA) (Figure 2a, b).^{14,35,40,58,59} The PARCELs were formulated at a ratio of 56/10/23/6/11 for SM-102/DOPE/cholesterol/C14-PEG-2000/polyphenol (Figure 2c, Figure S1d). The size, charge, and mRNA encapsulation efficiency for each PARCEL was reproducible, with sizes ranging from ~109 nm to ~154 nm (Figure 2d), PDI ranging from ~0.16 to ~0.28 (Figure 2e), zeta potentials ranging from ~−1.5 mV to ~1.0 mV (Figure 2f), and mRNA encapsulation efficiencies ranging from ~78.9% to ~92.4% (Figure 2g). It is noted that the size of PARCELs was similar to the LNP formulation and mRNA encapsulation efficiencies of PARCELs were greater than the LNP formulation. The pKa for each PARCEL was ranging from 6.8 to 7.4 (Figure S2).

Figure 2.

(a) Schematic illustration of mRNA-loaded PARCEL formulation via microfluidic chip. (b) Chemical structures of representative molecular excipients within mRNA-loaded PARCEL were used in this study. (c) Composition ratios for the formulation of mRNA-loaded PARCEL including GA, CAT, EGCG, and TA with the same weight ratio. (d) Size/Diameter, (e) PDI, (f) zeta potentials, and (g) mRNA encapsulation efficiency of PARCELs., (****p < 0.0001 and ***p < 0.001 with 95% of confidence level from unpaired t-test). In vitro FLuc expression of PARCELs treated on (h) B16-F10 and (i) DC 2.4 cells under 50, 100, and 200 ng mRNA dose per well across desired time (2, 4, 24, 48 h). In vitro EPO expression of PARCEL treated on (j) B16-F10 and (k) DC 2.4 cells under 50, 100, and 200 ng doses per well for 24 h. (All data presented as mean ± SD, n = 3).

Having evaluated the structure and the formulation of PARCELs, we then evaluated their in vitro efficacy by evaluating the protein expression in a dose-responsive (50, 100 and 200 ng) and a time-dependent (2, 4, 24 and 48 h) fashion using mRNA encoding for either firefly luciferase (FLuc, an intracellular protein) or human erythropoietin (EPO, a secreted protein) (Figure 2h–k, Table S2). Given the utility of mRNA therapies in cancer immunotherapy, we performed these studies on DC 2.4 cells (a dendritic cell line relevant as antigen presenting cells) and B16-F10 cells (a melanoma cell line). Upon collectively analyzing these data, several trends emerged. First, PARCELs were well-tolerated under each studied condition (Figure S3, S4). Second, FLuc and EPO expressions for each PARCEL were higher for DC 2.4 cells than for B16-F10 cells. Third, in the time-dependent scenario, FLuc expression for each PARCEL increased from 2 h to 24 h followed by a decrease in expression at 48 h (Figure 2h, i). Fourth, in a dose-responsive scenario, FLuc expression generally increased with FLuc mRNA dose from 50 ng to 200 ng (Figure 2h, i). Alternatively, treatment of B16-F10 cells and DC 2.4 cells with EPO mRNA PARCELs for 24 hours showed the highest EPO expression at 100 ng and 200 ng overall mRNA doses, respectively (Figure 2j, k). Finally, different PARCELs resulted in different levels of in vitro protein expression across both cell lines, highlighting the importance of polyphenol selection in PARCEL.

To explore the reasons for differences in protein expression, we next sought to explore several mechanistic studies to better understand mRNA delivery using each PARCEL. To begin, we quantified the cellular uptake of PARCELs in vitro using flow cytometry and confocal microscopy (Figure 3a–f, Figure S5, Table S3). To further complement these studies, we also sought to elucidate the specific endocytosis pathway and degradation properties of each PARCEL. In brief, these mechanism studies were performed by the inhibition of various endocytic pathways, specifically caveolin-mediated endocytosis, clathrin-dependent endocytosis, micropinocytosis, phagocytosis, and energy-dependent endocytosis (Figure 3g).^60–62 In collectively analyzing these data, several trends emerged. First, TA PARCEL had lower cellular uptake compared to other PARCELs across multiple time points (Figure 3a, c). Second, the cellular uptake of PARCELs was time-dependent, and maximum uptake was observed at 24 h for B16-F10 cells and 4 h for DC 2.4 cells. Third, each PARCEL was degraded after 24 h in B16-F10 cells and 4 h in DC 2.4 cells (Figure 3a, c), as indicated through the decrease in geometric mean fluorescence intensity (GMFI) after 24 h for B16-F10 cells and 4 h for DC 2.4 cells using flow cytometry (Figure 3b, d) and confocal microscopy (Figure 3e, f). Finally, phagocytosis was shown to be the main mechanism for non-TA PARCEL uptake, while micropinocytosis and energy-dependent endocytosis were important mechanisms for TA PARCELs (Figure 3g, h). Taken in tandem, these results suggest that PARCELs are internalized in a time-dependent fashion, through a combination of endocytic pathways for respective PARCELs.

Figure 3.

(a) Cellular uptake and (b) GMFI of B16-F10 cells treated with PARCEL at varying incubation times of 2, 4, 24, and 48 h. (c) Cellular uptake and (d) GMFI of DC 2.4 cells treated with PARCEL at varying incubation times of 2, 4, 24, and 48 h. Representative confocal microscopy images showing the intracellular trafficking of PARCEL in (e) B16-F10 and (f) DC 2.4 cells at varying incubation times of 2, 4, 24, and 48 h. Green: ATTO-488 labeled PARCEL; blue: nuclei; red: cell membrane. Scale bars are 10 μm. (g) Schematic illustration of the cell internalization mechanisms with corresponding related inhibitors used. (h) Study of the cell internalization mechanism of PARCEL by monitoring the cellular uptake efficiency in the presence of different endocytic inhibitors. Cells were treated with 500 ng mL⁻¹ of PARCEL at 37 °C (All data presented as mean ± SD, n = 3).

Following cellular internalization/uptake studies, we sought to understand how well each PARCEL could escape endosomal trapping. In brief, endosomal escape studies were performed by incubating nuclei and endo/lysosome labeled DC 2.4 cells with ATTO-488 labeled FLuc mRNA PARCELs and performing confocal microscopy to analyze the colocalization of PARCELs [Figure 4a–c; In these confocal images, cell nuclei are blue, mRNA-loaded PARCEL are green, and endo/lysosomes are red; yellow (i.e. colocalization of the green and red signals) suggests that the mRNA PARCEL remain trapped in endosomes]. As a benchmark, confocal imaging was also performed on cells incubated with analogous LNPs. As a quantifiable endosomal escape metric for each PARCEL, the Pearson Coefficient Correlation (PCC) was also determined (where a PCC value of 0 indicates complete endosomal escape and a PCC value of 1 indicates no endosomal escape).^63,64 To provide further insight into the endosomal escape properties of each PARCEL, we also investigated the buffering capacity of each PARCEL given that buffering capacity may relate to endosomal escape (Figure 4d).⁶⁴ To further add depth to our understanding of endosomal escape, “label-free” approaches for each PARCEL were also investigated (Figure 4e–g). In brief, these “label-free” studies were performed using enzyme inhibition/brightfield imaging studies with bafilomycin A₁ (a molecule that inhibits proton sponge aided endosomal escape by inhibiting V-ATPases)^65,66 and calcein (a membrane-impermeable dye that remains entrapped within intact endosomes but becomes distributed throughout cells if endo/lysosomes are ruptured), in which the calcein was directly added to the cells followed by adding the PARCEL or the bafilomycin A₁ (Figure 4e–g).

Figure 4.

(a) Schematic illustration of our endosomal escape studies using a lysotracker confocal imaging assay. (b) Representative confocal images of DC 2.4 cells treated with ATTO-488 labeled PARCEL (green). Endo/lysosomes (red) were stained with LysoTracker Deep Red. Nuclei (blue) were stained with Hoechst 33342. Scale bars are 10 μm. (c) Pearson Correlation Coefficient (PCC) analysis of ATTO-488 labeled PARCEL (Data presented as the mean ± SD, ***p < 0.001 and *p < 0.05 with 95% of confidence level from unpaired t-test, Figure S6). (d) Titration curves of PARCEL in suspensions as a function of HCl. Schematic illustration of our endosomal escape studies using (e) a calcein assay and (f) a proton sponge effect assay using bafilomycin A₁ for the termination of the inflow of H⁺ and Cl⁻. (g) Representative confocal images of DC 2.4 cells incubated with calcein and PARCEL in the absence (top row) and presence (bottom row) of inhibitor bafilomycin A₁ for 4 h at 37 °C. PARCELs were not fluorescently labeled to avoid interference with the calcein signal. Scale bars are 10 μm. (All data presented as mean ± SD, n = 3).

Upon analyzing these mechanistic data, several findings were observed. First, PARCELs had better endosomal escape than analogous LNPs as observed by lesser colocalization (i.e., less yellow color) in confocal microscopy images (Figure 4b) and lower PCC values (Figure 4c). Second, the pH value of CAT, EGCG, and TA PARCEL samples gradually decreased with the addition of HCl, as compared to the GA PARCEL, LNP, and MilliQ water (as a control), suggesting that different PARCEL can differentially buffer protons which may be important for endosomal escape (Figure 4d). Third, diffuse fluorescence of calcein dye was observed when cells were incubated with PARCEL compared to LNP (upper row, Figure 4g) further suggesting the superior endosomal escape properties of PARCELs as compared to analogous mRNA LNPs. Fourth, punctuated fluorescence was observed in cells treated with bafilomycin A₁ and PARCELs (bottom row, Figure 4g), suggesting that the ‘proton sponge effect’ could potentially be one of the mechanisms for triggering endosomal escape of PARCEL. Taken in tandem, these results highlight PARCELs as versatile mRNA carriers with tunable endosomal escape properties.

Building on the previous data, we finally sought to establish the in vivo delivery properties of each PARCEL. Briefly, Black 6 mice were treated with each PARCEL delivering mRNA encoding for FLuc (Figure 5a–c, Table S4) or EPO (Figure 5d, Table S5) via intravenous (IV) administration. Tolerability studies including histological evaluation (Figure 5e), liver and kidney function blood tests (Figure 5f) within complete blood paneling (Figure S7, S8), and weight loss studies (Figure S9, S10) were also evaluated for each PARCEL. Upon analyzing these data, several trends were observed. First, EGCG and TA PARCELs displayed higher FLuc expression than LNP as suggested by the increased FLuc signal in comparative IVIS imaging on the resected organs of treated mice (Figure 5a, c). Second, the increases in FLuc expression occurred without altering the innate biodistribution of each studied PARCEL (Figure 5b). Third, EGCG and TA PARCELs also increased the amount of EPO expression secreted into the blood of treated mice (Figure 5d). Fourth, each PARCEL was well tolerated as analyzed by histology (Figure 5e), weight retention (Figure S9, S10), and complete blood paneling data including normal alkaline phosphatase (ALP), alanine transaminase (ALT), aspartate transferase (AST), blood urea nitrogen (BUN) and creatinine (CREAT) levels which are markers of liver and kidney function (Figure 5f). Taken in tandem, these results suggest that EGCG and TA PARCEL had better in vivo protein expression than analogous LNP formulations, highlighting their potential for therapeutic mRNA delivery.

Figure 5.

(a) Representative luminescence biodistribution of PARCEL encapsulated with FLuc mRNA ex vivo (n = 3) for each group via intravenous injection. Mice injected with naked FLuc mRNA and PBS were used as controls. (b) Associated percent of bioluminescence and (c) total luminescence of FLuc mRNA encapsulated PARCEL across various organs including the pancreas, spleen, liver, kidneys, uterus/ovaries, lung, and heart. (d) Human EPO concentration after the injection of EPO mRNA encapsulated PARCEL for 24 h. Mice injected with naked EPO mRNA and PBS were used as controls. The concentration of human erythropoietin was characterized by Human EPO ELISA kits following the manufacturer’s protocol. (e) Representative histology images of the liver, spleen, and lung of mice after treatment with FLuc mRNA encapsulated PARCEL via IV injection routes (n = 3). Scale bars are 50 μm. (f) ALP, ALT, AST, BUN, and CREAT blood testing results after the IV injection of FLuc mRNA encapsulated PARCEL (ns > 0.05 with 95% confidence level from unpaired t-test with PBS group, and all data presented as mean ± SD, n = 3).

In this report, we provide computationally guided, formulation-driven, and mechanism-driven studies to realize the development of PARCELs as a safe and effective mRNA delivery platform. In brief, we demonstrate their effectiveness as an mRNA delivery system by evaluating their physiochemical properties including the size, PDI, charge, encapsulation efficiency, as well as the mechanisms behind their cellular performance such as endocytosis and endosomal escape. Furthermore, our research also showed that TA PARCEL exhibited the best in vivo efficacy on both intracellular and secreted protein expression. Future work will be directed toward assessing the utility of PARCEL in the field of cancer immunotherapies and furthering the therapeutic utility of PARCEL for mRNA delivery. Taken collectively, the goal of this study was to establish PARCEL as a viable platform for mRNA delivery with superior endosomal escape properties to analogous LNPs while more broadly highlighting the utility of synergizing techniques in structural analysis, formulation, and mechanism to afford better therapies for the study and prevention of disease using mRNA.