Comparative optimization of polysaccharide-based nanoformulations for cardiac RNAi therapy

Abstract

Ionotropic gelation is widely used to fabricate targeting nanoparticles (NPs) with polysaccharides, leveraging their recognition by specific lectins. Despite the fabrication scheme simply involves self-assembly of differently charged components in a straightforward manner, the identification of a potent combinatory formulation is usually limited by structural diversity in compound collections and trivial screen process, imposing crucial challenges for efficient formulation design and optimization. Herein, we report a diversity-oriented combinatory formulation screen scheme to identify potent gene delivery cargo in the context of precision cardiac therapy. Distinct categories of cationic compounds are tested to construct RNA delivery system with an ionic polysaccharide framework, utilizing a high-throughput microfluidics workstation coupled with streamlined NPs characterization system in an automatic, step-wise manner. Sequential computational aided interpretation provides insights in formulation optimization in a broader scenario, highlighting the usefulness of compound library diversity. As a result, the out-of-bag NPs, termed as GluCARDIA NPs, are utilized for loading therapeutic RNA to ameliorate cardiac reperfusion damages and promote the long-term prognosis. Overall, this work presents a generalizable formulation design strategy for polysaccharides, offering design principles for combinatory formulation screen and insights for efficient formulation identification and optimization.

Fig. 1. Schematic illustration of workflow for the current study.

The current research arose with the construction of cationic compounds collections with distinct categories, and a previously synthesized multi-functional polysaccharide was fixed as ionic part. The screen process was aided by a high-throughput microfluidics workstation coupled with streamlined NPs characterization system, and the step-wise screen was interpreted by a computational aided analysis for potentially understanding the decision-making mechanism. In the end, the out-of-bag RNAi nanomedicine was tested in a murine myocardial reperfusion model to evaluate its corresponding therapeutic potency.

Fig. 2. “Robot assisted” microfluidic preparation and screening of EEPG-paired nanoparticles.

a Overview of the self-designed automated microfluidics system for formulation preparation; b Chemical structures and names of 20 cationic compounds for constructing EEPG-based NPs in a combinatorial manner. C1-C4, Category 1-4; c Graphic illustration of the initial screening pool, where each category of cationic compounds was set 7 different weight ratios (w/w) or nitrogen/phosphate (N/P) ratios, and blue dot indicated the formation of NPs, whereas grey dot indicated the weak interaction of the two compartments without pervasive formation of NPs; d Representative nanoformulations with optimal hydrodynamic diameter and polydispersity (PDI) (n = 3, replicates), e Zeta (ζ)-potential value of representative 16 nanoformulations (n = 3, replicates). Data are presented as mean ± SD, n = 3. Source data are provided as a Source Data file.

Fig. 3. Step-wise screening of EEPG-paired nanoparticles for potential siRNA delivery.

RiboGreen assay was adopted to determine the encapsulation efficiency of siRNA in each cationic compound-paired nanoparticles, the nitrogen/phosphate (N/P) ratios were set at 15/1 (a) and 20/1 (b) respectively; c THP-1 cytotoxicity evaluation of 7 nanosystems including spermine-EEPG, PEI800-EEPG, PEI25K-EEPG, Lipid 2-EEPG, Lipid 10-EEPG, K9-EEPG and KALA-EEPG, the concentrations were correspondence to the total NPs concentration; d, e Western blotting analysis of GAPDH expression in THP-1 derived macrophages. α-Tubulin was used as an internal control. Data are presented as mean ± SD with triplicates (n = 3). Source data are provided as a Source Data file.

Fig. 4. Computational assisted step-wise interpretation of formulation screening process.

a A decision tree-like scheme was built up to summarize the screening process, where dark blue squares represent the positive sets identified at each stage of screening, while grey squares denote the negative sets. Additionally, light blue diamonds are used to delineate the specific criteria applied at each screening juncture. KD (knockdown); b A summarized table to show the tested cationic compounds in the computational analysis; c Quantitative analysis of Coulombs potential by 1 ns molecular dynamics (MD) simulations with different EEPG pairings; d Quantitative analysis of average inner Lennard-Jones (LJ) potential from EEPG by 1 ns molecular dynamics (MD) simulations from cationic compounds that did not form stable NPs with EEPG at any concentration (Negative Samples, samples 15, 16, 17, and 20), and cationic compounds formed colloidal stable NPs only at N/P ratios significantly deviating from unity (Non-Electronic Binding, samples 14, 18, and 19). Data are presented as mean ± SD. Statistical analysis was conducted by two-tailed Student’s t-test, where *p = 0.0248; e Linear correlation between encapsulation efficiency and average LJ potential between EEPG and the paired cationic compound; f Molecular dynamics (MD) simulation was performed to reflect the binding structure of Transportan-EEPG and KALA-EEPG; g Coefficient of determination of different MolDes in correlation to IC50 of each formulation as calculated by feature selection. Blue circles indicated the top 5 MolDes with the highest Pearsons’ correlation coefficient and light circles indicated MolDes which were previously reported to be highly correlated to the cytotoxicity of a compound; h Linear correlation between IC50 of the formulation and qnmax value from cationic compound; i Squared Pearsons’ correlation coefficient of top 5 MolDes in correlation to knockdown efficiency of each formulation; j Linear correlation between knockdown efficiency of the formulation and TDB10p value from cationic compound. Grey area in e, h and j represent 95% confidence interval. Source data are provided as a Source Data file.

Fig. 5. Characterization and in vitro evaluation of GluCARDIA NPs.

Morphology of bare GluCARDIA NPs (a) and GluCARDIA-siRNA NPs (b) were observed by TEM analysis; c DLS analysis of GluCARDIA NPs and GluCARDIA-siRNA NPs to evaluate their corresponding hydrodynamic diameter and PDI (n = 3, replicates); d Time-dependent confocal microscopy images of GluCARDIA-Cy5 siRNA in THP-1 macrophages delineating the intracellular translocation of Cy5-siRNA; e Mander’s Correlation Coefficient (MCC) between Cy5-siRNA and Lyso-Tracker at different time-points were quantified to show the time-dependent endosome escape of siRNA (n = 5); f Flow cytometry results of GluCARDIA Cy5-siRNA NPs uptake in RAW 264.7 and THP-1 cells (n = 3, replicates); Western blotting images (g) and corresponding quantitative analysis of GAPDH expression in THP-1 or RAW 264.7 cells after treated with GluCARDIA-siGAPDH NPs at different siRNA concentrations (h), (n = 3, replicates). α-Tubulin was used as an internal control. Data are presented as mean ± SD with at least triplicates. Statistical analysis was conducted by two-tailed Student’s t-test, where *p < 0.05; **p < 0.01; ***p < 0.001. Source data are provided as a Source Data file.

Fig. 6. Biodistribution and gene silencing efficacy of GluCARDIA NPs in murine cardiac IR model.

a–c Short term liver toxicity from GluCARDIA NPs upon intravenous injection were evaluated by serum alanine aminotransferase (ALT), aspartate aminotransferase (AST) and total bilirubin (TBIL) levels (n = 3, replicates); d Representative fluorescent mapping of ICG-Dextran NPs and ICG-GluCARDIA NPs in major organs at 24 h post-IR. e Heart targeting index (HTI) was quantified to compare the heart accumulation efficacy of ICG-Dextran NPs and ICG-GluCARDIA NPs (n = 3, replicates). f–g Representative dual-immunofluorescence staining of ICG-GluCARDIA NPs and Dectin-1 (f) or ICG-GluCARDIA NPs and Ly-6G (g) at 24 h post-IR; h, Mander’s Correlation Coefficient (MCC) between ICG-GluCARDIA and Dectin-1 or ICG-GluCARDIA and Ly6G were quantified to confirm the cellular interaction of GluCARDIA NPs (n = 4, replicates); i–k Western blotting analysis of IRF3 and pIRF3 expression in injured heart tissues. α-Tubulin was used as an internal control (n = 3, replicates). Data are presented as mean ± SD with at least triplicates. Statistical analysis was conducted by two-tailed Student’s t-test or one-way ANOVA with Tukey’s correction, where *p < 0.05; **p < 0.01; ***p < 0.001 and ****p < 0.0001. Source data are provided as a Source Data file.

Fig. 7. GluCARDIA-siIRF3 NPs ameliorated cardiac IR prognosis.

a Schematic illustration of experimental strategy. Mice established with IR injury were intravenously (i.v.) injected with GluCARDIA-siIRF3 at 30 min, 24 h and 72 post-reperfusion; b Gene Set Enrichment Analysis (GSEA) between IR and GluCARDIA-siIRF3 at day 7 post-reperfusion indicting a reduced TGF-β signaling pathway activation from GluCARDIA-siIRF3 group, potentially indicating a less fibrosis extent; c Representative echocardiographic images at day 7 post-IR; d Day 7 echocardiography-based quantification of left ventricular end-systolic internal dimension (LVIDs), left ventricular end-diastolic internal dimension (LVIDd) and ejection fraction (EF) from different groups (n = 8); e Representative echocardiographic images on day 28; f Day 28 echocardiography-based quantification of LVIDs, LVIDd and EF from different groups (n = 8); g Representative Masson’s Trichrome staining of heart tissue sections from different groups at day 28 post-reperfusion; h Masson’s Trichrome staining-based quantification of left ventricular (LV) wall thickness from different groups (n = 5); i Masson’s Trichrome staining-based quantification of fibrotic tissue percentage from different groups (n = 5); j Representative images of wheat germ agglutinin (WGA) staining to visualize cardiomyocytes in sections from remote area (n = 5), scale bar: 10 μm. Data are presented as mean ± SD with at least triplicates. Statistical analysis was conducted by two-tailed Student’s t-test, where *p < 0.05; **p < 0.01; ***p < 0.001 and ****p < 0.0001. a Created with BioRender.com. Source data are provided as a Source Data file.