Lipid nanoparticle formulation for gene editing and RNA-based therapies for glioblastoma

Abstract

Background

Glioblastoma (GBM), one of the deadliest cancers, resists current therapies, with drug development hindered by its high heterogeneity. However, GBM consistently relies on microRNA-10b (miR-10b), a key driver of glioma growth and a promising therapeutic target. miR-10b gene editing represents a potential treatment, but effective delivery strategies for gene editing systems in GBM remain unexplored.

Methods

We developed lipid nanoparticles (LNPs) encapsulating Cas9 mRNA and a miR-10b-targeting sgRNA (termed miRTEN). miRTEN was tested in glioma stem cells (GSCs) and orthotopic GBM models to assess therapeutic efficacy, immune responses, and safety.

Results

Intracerebroventricular injections of miRTEN enabled broad and durable Cas9 mRNA expression and miR-10b gene editing in tumor core and invasive areas across diverse GBM models. miRTEN significantly suppressed tumor growth, reduced GSC proliferation and viability, with therapeutic outcomes correlating with dose-dependent miR-10b suppression. Combining miRTEN with temozolomide (TMZ) further enhanced tumor suppression, overcoming TMZ resistance and improving survival. In immunocompetent models, miRTEN activated antitumor immune responses, increased cytotoxic CD8⁺ T cells infiltration, and promoted durable immune memory, enabling tumor rejection upon rechallenge. Safety assessments demonstrated that miRTEN selectively targets GBM cells, sparing normal brain tissues and causing no significant off-target toxicity.

Conclusion

As in vivo CRISPR-based drugs advance toward clinical applications, our findings demonstrate the potential of LNPs-mediated CRISPR–Cas9 systems for targeted miR-10b editing and, more generally, gene editing and RNA therapies for GBM. miRTEN monotherapy, as well as its combination with standard care, offers a promising, safe, and effective approach to improving outcomes in GBM.

Graphical Abstract

Graphical Abstract.

Key Points.

• Safe and scalable LNPs enable in vivo gene editing and RNA therapies for GBM.

• miR-10b editing inhibits GBM growth through cell-autonomous, bystander, and immune mechanisms.

• The drug can be administered as a monotherapy or in combination with SOC.

Importance of the Study.

Targeted delivery of gene and RNA therapies for Glioblastoma (GBM) remains largely unexplored. Here we present biodegradable RNA-encapsulating lipid nanoparticles (LNPs) that enable broad, tumor-enriched expression in orthotopic GBM models. Intracerebroventricular (ICV) administration of these LNPs, formulated with Cas9 mRNA and sgRNAs, achieves efficient gene editing within the tumor.

We validate this approach by targeting miR-10b, a key onco-miRNA commonly expressed in GBM and essential for GBM growth. miR-10b-editing nanoparticles (designated miRTEN) significantly suppress tumor growth, activate robust antitumor immunity, and improve survival while overcoming temozolomide resistance. Remarkably, miRTEN induces durable immune memory and exhibits a promising safety profile.

The drug formulation is safe and scalable, leveraging LNP technologies already validated in mass production and clinical applications. Thus, it provides a versatile platform for CRIPSR and RNA replacement therapies, offering new avenues for treating GBM and other high-grade gliomas with unmet clinical needs.

GBM is a highly aggressive brain tumor with poor prognosis—median survival remains ~15 months despite maximal surgical resection, radiotherapy, and temozolomide (TMZ) chemotherapy.^1,2 Novel strategies targeting GBM’s molecular drivers are urgently needed. A major obstacle to effective treatment is the persistence of glioma stem cells (GSCs), which drive therapy resistance, infiltration, and recurrence.³ Another key challenge is the disease’s high heterogeneity, marked by multiple mutations and dysregulated signaling pathways. As a result, targeted therapies typically address only subsets of GBM with specific mutations, such as EGFR or IDH1 pathogenic variants.^1,2 In contrast, microRNA-10b (miR-10b) has emerged as a broadly relevant therapeutic target: it is consistently upregulated in malignant gliomas, absent in normal brain, and essential for glioma growth and survival.^4-7 Its inhibition could offer widespread therapeutic benefit to GBM patients, bypassing the need for tumor stratification.

miR-10b is a critical oncogenic regulator in GBM, promoting tumor-cell proliferation, invasion, and apoptosis resistance.^4-11 It regulates critical cancer genes including CDKN1A/p21, CDKN2A/p16, BCL2L11/Bim, which typically protect cells from uncontrolled growth, via its miRNA-mediated mechanism.^4-6 Beyond this, miR-10b exerts a unique nuclear role by binding the core spliceosomal snRNA U6, disrupting its function and altering splicing of numerous genes implicated in gliomagenesis.⁷ These multifaceted actions make miR-10b inhibition a promising therapeutic approach, achievable via antisense oligonucleotides or gene editing.^5,12,13 Lentiviral CRISPR–Cas9-mediated miR-10b gene editing (GE) demonstrated strict glioma cell dependence on miR-10b, highlighting the therapeutic potential of this strategy.¹² Additional key benefits of miR-10b GE include its selectivity, due to miR-10b’s expression and chromatin accessibility in GBM but not normal cells of the brain, and the disruption of its short target sequence, where even minor alterations (in either 5′ or 3′ miR-10b ends, or adjacent sequences) impair its function and/or maturation.¹²

However, GBM’s intracranial location, diffuse growth, and infiltrative nature pose a challenge to drug delivery.¹⁴ As a proof-of-principle for miR-10b GE, we employed lentiviral vehicles¹²; however, integrating lentivectors are not considered suitable for in vivo use due to the risk of insertional mutagenesis. Lipid nanoparticles (LNPs) have emerged as effective RNA therapeutic carriers, given their ability to encapsulate and protect RNA molecules while enhancing cellular uptake.¹⁵ For GE, LNPs may have several major advantages over virus-mediated CRISPR/Cas9 delivery,^16,17 most notably the ability to repeat doses to patients—something currently not possible with viral vectors. Additional advantages include the potential to encapsulate larger RNA cargo (such as Cas9 mRNA), provide broad intratumoral distribution, and enabling transient CRISPR/Cas9 expression to limit nuclease exposure and reduce toxic off-target effects. Using clinically tested ionizable cationic lipids, we established LNP formulations for efficient and durable GE in intracranial GBM. This approach selectively disrupts miR-10b’s oncogenic activity and inhibits tumor growth, demonstrating LNPs’ promise as a platform for advancing gene editing and, more broadly, RNA-based therapies, potentially improving GBM treatment outcomes.

Materials and Methods

LNP Formulation

LNPs were optimized based on Finn et al.¹⁸ with an amine-to-RNA-phosphate (N:P) ratio of 4.5. Lipid components were dissolved in ethanol at molar ratios: 45% ionizable lipids (LP01 or DLin-MC3-DMA or SM-102), 44% cholesterol, 9% DSPC, and 2% PEG2k-DMG. The RNA cargo (1:1 weight ratio mRNA:sgRNA) was dissolved in 50 mM acetate buffer (pH 4.0). For in vitro experiments, LNPs were generated using hand-mixing, while microfluidic mixing (using Precision NanoAssemblr Ignite + Instrument, flow rate ratio 3:1; total flow rate 12 mL/min) was used for in vivo experiments. Encapsulation efficiencies were determined by Ribogreen assay (Invitrogen, R11490). Particle size and polydispersity were measured by dynamic light scattering. CleanCap^® Cas9 mRNA (polyadenylated, 5-methoxyuridine modified) was purchased from TriLink (L-7206) and modified sgRNAs¹⁸ from AXO Labs. sgRNA sequences are listed in Supplementary Table 1.

Cell Cultures and LNP Treatments

LN229 and CT2A glioma cells were maintained in DMEM (Gibco) supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin. Human low-passage GBM8, GBM6, GBM4 (gifts from Dr. Hiroaki Wakamoto), GBM30, GBM34, GBM44, GBM62, GBM88, GBM146, GBM333, and GBM816 cells (gifts from Dr. Pier Paolo Peruzzi),¹⁹ and GBM39 (gift from Dr. Alain Charest), were cultured as neurospheres in serum-free Neurobasal medium. For LNP treatments, glioma adherent cells (5000 per well) were plated in 96-well plates, followed by LNPs addition and incubation for 3 to 5 days. GSC spheroids were dissociated into single cells before LNP treatment. LNPs were preincubated with 1.5 µg/ml ApoE (R&D Systems, 4144-AE) for 5 min at 37°C and used at doses ranging from 25 to 200 ng Cas9 mRNA/sgRNAs per well. A dose of 100 ng/well was defined as low and 200 ng/well as high. Rat primary neuroglial cells were derived from E18 embryos and cultured on Poly-D-Lysine-coated 96-well plates.²⁰ Human iPSC-derived neurons were differentiated using a doxycycline-inducible NGN2 expression system²¹ and treated with LNPs at DIV14.

Quantitative Real‐time Reverse Transcriptase PCR

RNA was extracted using the Total RNA Purification Kit (Norgen Biotek, 17250) and treated with DNase I (Norgen Biotek, 25720). For miRNA analysis, 200 ng RNA was reverse-transcribed to cDNA using the miRCURY LNA RT kit (QIAGEN, 339340), followed by qPCR with miRCURY LNA SYBR Green PCR kit (QIAGEN, 339347) using specific primers (miR-10b, miR-125a-5p, miR-99a-5p; QIAGEN). For mRNA analysis, 200 ng RNA was reverse-transcribed with iScript cDNA Synthesis Kit (BIO-RAD, 1708841), and qPCR was performed using SYBR Green assay (Thermo Fisher, 4367659) with primers (Supplementary Table 2). Expression was normalized to miR-125a-5p (miRNAs) or GAPDH (mRNAs) via 2^−ΔΔCt method.

DNA Sequencing

DNA was extracted using the DNA RapidLyse kit (Macherey-Nagel, 740100.50). miR-10b PCR amplicons were generated using Phire Green Hot Start II PCR Master Mix (Thermo Scientific, F126L) with specific primers (forward 5′-GTGGTGGCTCAGAGGAAGAG-3′ (for human), 5′-GGTGGCTGTGCTGAAGAGAT-3′ (for mouse), and reverse 5′-TGAAGTTTTTGCATCGACCA-3′ (for both human and mouse)). PCR products were purified using the Monarch^® PCR & DNA Cleanup Kit (New England Biolabs, T1030L). CRISPR–Cas9-induced mutations were detected by Next-Generation Sequencing with editing efficiency analyzed using CRISPResso2 (http://crispresso2.pinellolab.org/submission).

Cell Viability Assessment and Phenotypic Analysis

Glioma cells viability was assessed using WST-1 assay (Roche, 11644807001). GSCs viability was measured with CellTiter-Glo^® 2.0 assay (Promega, G9242), while neuronal viability was tested using AlamarBlue reagent (Invitrogen, A50100). GSC spheroid size was assessed using a GE INCELL Analyzer 2200 and ImageJ. Neuronal morphology and neurite length were analyzed using the Incucyte S3 Live-Cell Analysis Instrument and NeuroTrack module.

Orthotopic GBM Models

Orthotopic GBM tumors were established as previously described,^5,12 by injecting mCherry-Fluc-labeled GSCs or CT2A cells into the striatum of 6- to 8-week-old female C57BL/6 or nu/nu mice. LNPs were administered via bilateral ICV injections (5 µL per side, 10 to 30 µg per mouse) at 500 nL/min, with needle placement at 0.2 mm AP, ±0.8 mm ML, and 3.0 mm DV from bregma. TMZ was given at 50 mg/kg every other day for 5 doses. Tumor growth was monitored using In Vivo Imaging System. See Supplementary Methods for details.

Immunofluorescence

Cells were cultured on Poly-D-Lysine-coated surfaces, treated, and fixed with 4% paraformaldehyde (PFA). Frozen GBM tissues were fixed, dehydrated in sucrose, embedded in OCT, sectioned, and acetone-treated. Paraffin-embedded sections were deparaffinized, rehydrated and subjected to antigen retrieval. Permeabilization was performed with 0.5% Triton-X 100, followed by blocking with 5% horse serum (Gibco). Primary antibodies were incubated overnight at 4°C, followed by secondary antibodies incubation at room temperature for 2 h and DAPI staining. See Supplementary Methods for details.

Flow Cytometry Analysis of Cell Death

GSCs were dissociated into single-cell suspensions using the NeuroCult™ Chemical Dissociation Kit (STEMCELL Technologies, 05707), washed twice with cold PBS and stained with 7-AAD viability dye (BD Biosciences) at room temperature for 15 min in the dark. Samples were then analyzed on a BD LSRFortessa™ flow cytometer, and data were processed using FlowJo software.

Analysis of Edited CD8⁺ T Cells

Naïve CD8⁺ T cells were isolated from the spleens of 6- to 8-week-old C57BL/6 mice and purified using magnetic negative selection. Cells were activated for 12 h on 96-well plates coated with anti-CD3/CD28 (2 μg/mL each) and then treated with miRTEN or control LNPs (200 ng/well) for 24 h. The cells were then stimulated with recombinant IL-2 (10 ng/mL) and IL-12 (20 ng/mL) for 48 h to promote effector CD8 + T differentiation. To assess the effector function, cells were restimulated with phorbol-12-myristate 13-acetate (PMA, 50 ng/mL) and ionomycin (1 μg/mL) in the presence of Golgi Stop for 4 h. Viability was assessed using a fixable dye, followed by fixation, permeabilization, and intracellular staining for IFN-γ, Granzyme B, TNF-α and CD8. Flow cytometry was performed and data was analyzed based on fluorescence minus one (FMO) controls using FlowJo software (v10.10). All antibodies and reagents are listed in Supplementary Methods.

Neuronal Activity

Human iPSC-derived neurons were seeded at 30  000 cells/well in PEI/laminin-coated Greiner black-walled, clear-bottom 96-well plates. Synapsin-driven GCaMP6s was introduced via lentiviral transduction at MOI 10 on DIV19. Primary human astrocytes (10 000 cells/well) were added on DIV11. Baseline calcium imaging was performed at DIV31. Weekly LNP treatments (200 ng Cas9 mRNA + 200 ng sgRNA/well) began on DIV33, with half-medium changes at each treatment. Long-term calcium imaging was conducted on DIV39, DIV46, DIV55, and DIV61 followed by weekly live cell Incucyte imaging for cell viability. Three random regions of interest were recorded for 2.5 min each to extract GCaMP6s fluorescence from individual neurons. ΔF/F₀, calcium amplitude, and inter-spike interval (ISI) were quantified using CaSiAn analysis toolbox and compared across groups. See Supplementary Methods for details.

Statistical Analysis

Details are provided in the figure legends.

Results

LNP-formulated CRISPR–Cas9 System for Therapeutic Gene Editing in Malignant Glioma

To advance gene editing therapies for GBM and other malignant brain tumors, we developed LNPs encapsulating Cas9 mRNA and guide RNAs (sgRNAs) in the same particles. A critical LNP design variable is the ionizable lipid type and content, essential for RNA encapsulation and endosomal escape.²² We tested three clinically approved ionizable lipids—DLin-MC3-DMA (MC3), SM-102, and LP01—for their ability to co-deliver Cas9 mRNA and a GFP-targeting sgRNA in GSC (GFP⁺ GBM8). LP01-based LNPs outperformed others, showing high Cas9 expression and an 8-fold reduction in GFP versus 3-fold for MC3 and SM-102 (Supplementary Figure 1A). Next, we tested LNPs containing Cas9 mRNA and a miR-10b-targeting sgRNA (either G1 targeting mature miR-10b, or G3 targeting pri-miR-10b), previously validated in lentivirus-mediated experiments¹² (Figure 1A). LP01-based LNPs most effectively reduced glioma growth and viability across LN229 monolayers and GBM8 spheroids (Supplementary Figure 1B and C). These miR-10b-editing nanoparticles were designated miRTEN and selected for subsequent experiments.

Cas9 expression in glioma cells treated with miRTEN was dose-dependent: 100 and 200 ng/well doses induced expression in 30% to 70% and 80% to 100% of cells, respectively (Figure 1B). Corresponding targeted sequencing of miR-10b locus showed editing efficiencies (EEs) for both G1 and G3 rising with dose, from ~5% (25 ng) to 75%-85% (200 ng) (Figure 1C). Editing efficiency closely matched miR-10b suppression measured by qRT-PCR (Figure 1D), with strong correlation (R > 0.83, P < 0.001; Figure 1E). Both G1 and G3 miRTENs de-repressed known direct miR-10b targets (e.g. CDKNA1/P21, SART3, MBNL1, DGCR14, and snRNA U6),^5-7 and modulated downstream indirect factors (E2F1 and CDC42),^4,5,7 confirming effective functional inhibition of miR-10b (Figure 1F).

Figure 1.

LNP-mediated miR-10b GE suppresses miR-10b, modulates downstream signaling, and reduces viability of diverse glioma and GSC cells. (A) A schematic representation of alternative G1 and G3 sgRNAs used for CRISPR–Cas9 miR-10b editing. Mature miR-10b sequence is highlighted in blue. (B) Cas9 expression was analyzed in LN229 glioma cells and GBM8 spheroids treated with LNPs carrying either Cas9 mRNA and GFP sgRNA (control) or miR-10b-specific sgRNAs (G1 or G3). Cells received low (100 ng Cas9 mRNA and 100 ng sgRNA) or high (200 ng each) doses for 3 days (n = 3). Immunofluorescence demonstrated dose-dependent Cas9 distribution, with robust expression in treated cells. (C) miR-10b GE efficiency was assessed across various LNP concentrations in LN229 cells using targeted DNA sequencing (n = 3). Editing efficiency was analyzed using the CRISPResso2 tool, with sequencing reads aligned to a reference sequence, and the frequency of insertions, deletions, and substitutions quantified across the amplicon. (D) miR-10b levels were measured by qRT-PCR in LN229 and GBM8 cells treated with LNPs containing Cas9 mRNA and GFP sgRNA, or Cas9 mRNA with miR-10b-specific sgRNAs at low and high doses for 3 days (n = 3-5). (E) A correlation analysis between miR-10b editing efficiency and miR-10b levels in LN229 cells indicates an inverse relationship. (F) qRT-PCR analysis revealed that miR-10b GE in LN229 cells derepresses miR-10b targets and signaling. mRNA levels were normalized to GAPDH, and miR-125a served as the normalization control for U6 snRNA (n = 5). (G) miR-10b expression in multiple GSC lines and normal human neural cells was measured by qRT-PCR, normalized to miR-125a-5p (n = 3). (H) GSC cultures with varying levels of miR-10b were treated with low (100 ng) and high doses (200 ng) of miRTEN for 5 days (n = 5 samples). Representative images are shown. (I) GSCs treated with 100 ng of miRTEN (G1 or G3) or control LNPs (with Cas9 mRNA and GFP-sgRNA) for 3 days were stained for Ki67 (n = 3). Representative images and quantification of Ki67-positive cells are shown. (J) GSCs treated with 200 ng miRTEN (G1 or G3) or control LNPs for 5 days were analyzed for cell death via 7-AAD flow cytometry staining. Representative plots and 7-AAD⁺ cell percentages are shown (n = 3). All data are represented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Dunnett’s multiple comparison. ns, no significance; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

miR-10b Gene Editing Reduces Viability of Diverse GSCs

GBM is a highly heterogeneous disease. To assess whether LNP delivery varies across GBM molecular subtypes, we treated patient-derived GSCs with DiD-labeled LNPs and evaluated uptake by immunofluorescence after 48 h. GSCs representing proneural (GBM8, GBM44), mesenchymal (GBM34, GBM62), and classical (GBM6, GBM39) subtypes all showed comparable uptake levels (80%-90% DiD-positive cells) and Cas9 expression, suggesting consistent delivery across transcriptional subtypes (Supplementary Figure 2).

Although miR-10b is broadly expressed in glioma cells and absent in normal neuroglial cells—making it a unique glioma marker^5,6—its expression levels varies across different glioma models.⁵ To determine whether miR-10b GE reduces growth and viability across various GSCs, we selected patient-derived GSCs with low miR-10b levels and compared them to cultures with typical high miR-10b expression (Figure 1G). We tested miRTEN’s effects in high-miR-10b (GBM8, GBM44, GBM816) and low miR-10b (GBM34, GBM62, GBM88) GSCs. Treatment produced dose-dependent reductions in viability: 40%-60% in low-miR-10b and 70%-95% in high-miR-10b GSCs at 200 ng miRTEN. Both G1 and G3 formulations yielded comparable inhibitory effects across GSC subtypes (Supplementary Figure 3A). Sphere growth, including spheroid size and number, declined in all GSC cultures (Figure 1H; quantification in Supplementary Figure 3B). A short (72 h), low-dose treatment strongly reduced %Ki67 cells, confirming miR-10b’s role in glioma proliferation (Figure 1I; Supplementary Figure 3C). 7-AAD+ staining showed increased cell death: 12%-20% with low-dose and 35%-45% with high-dose treatment after 5 days (Figure 1J). Overall, while growth inhibition was stronger in high-miR-10b cells, miRTENs effectively reduced viability across all tested glioma models.

Previous studies showed that lentiviral miR-10b editing induces a strong bystander effect that enhances glioma cell death.²³ Similarly, we found that conditioned media (CM) from miRTEN-edited cells reduced growth and viability of naïve glioma cells and GSCs (Supplementary Figure 4A and B). CM from edited GSCs inhibited spheroid growth in naive GBM8 and GBM62 cells, reducing both spheroid size and number (Supplementary Figure 4B). These findings suggest that high-efficiency miR-10b editing may not be required to achieve therapeutic efficacy in vivo.

Broad and Durable Cas9 Expression in Diverse Orthotopic GBM Models

For in vivo orthotopic GBM studies aimed at clinical translation, we developed a scalable protocol to formulate miRTEN using microfluidic mixing of lipid and RNA solutions via Precision NanoAssemblr Ignite + system. This method consistently produced particles of 100-120 nm with high RNA encapsulation efficiency (>90%) (Supplementary Figure 5A and B). We first formulated GFP mRNA into LNPs and administered them via bilateral intracerebroventricular (ICV) injections to established orthotopic mCherry⁺ GBM models. In both a fast-growing, invasive GBM8 model and a slower, less invasive GBM4 model, GFP expression was robust and widespread across tumors post-LNP administration (Supplementary Figure 6A and B). We then evaluated distribution and expression of Cas9 (encoded by a much longer mRNA) in 3 different orthotopic GBM models: 2 human GSC-based xenografts (GBM8 and GBM4), and the mouse CT2A allograft in immunocompetent mice. A single bilateral ICV injection of total 10 µg of LNPs carrying Cas9 mRNA achieved broad distribution and high Cas9 expression in GBM tumors, with approximately 40% of tumor cells expressing Cas9 at 48 h post-injection, increasing to over 60 %-70% by 72 h post-injection in GBM8 model (Figure 2A). Similarly, about 50%-60% of tumor cells in both GBM4 and CT2A models expressed Cas9 at 72 h post-injection (Figure 2B and C). Cas9 expression was prominent in both the tumor core and invasive edges, while expression in surrounding normal brain tissues remained low (5%-10%) in both nude and immunocompetent mice (Figure 2A and C). Notably, broadly expressed Cas9 was detected throughout GBM tissues (in over 50%-70% of tumor cells) up to 3 weeks after injection, demonstrating that LNP administration leads to prolonged Cas9 expression (Figure 2D).

Figure 2.

Broad and durable Cas9 expression in orthotopic GBM models. (A-C) human GBM8 (1 × 10⁵), GBM4 (1 × 10⁵), and mouse CT2A (5 × 10⁴) cells expressing FLuc and mCherry were stereotactically implanted into the brains of nu/nu or C57BL/6 mice. Single bilateral ICV injections (5 µl/side) of LNPs containing 10 µg Cas9 mRNA were administered to established tumors at 30-, 50-, and 20-days post-implantation, respectively. Control mice received PBS. Cas9 biodistribution was analyzed 48-72 h post-injection in whole-brain sections, with strong Cas9 expression in tumor cores and invasion areas, and minimal expression in normal brain parenchyma. (D) Broad Cas9 expression in GBM8 tumors was detected at 1 and 3 weeks post-treatment, with representative images shown. (E) GFP⁺ GBM8 cells (1 × 10⁵) were implanted similarly, and LNPs with Cas9 mRNA, or Cas9 mRNA/GFP sgRNA (10 µg/mouse, n = 3/group) were injected post-establishment. GFP fluorescence was significantly reduced 5 days post-injection (ZEN software). Statistical significance was determined using Student’s t test. **P < 0.001.

We further evaluated the in vivo GE efficacy in orthotopic GBM8 tumors. GFP-expressing tumors were established and treated with 10 µg of LNPs encapsulating Cas9 mRNA and GFP sgRNA via ICV injections. Five days after injection, whole-brain analysis showed approximately 65% reduction in GFP expression in tumors treated with the GFP-editing formulation, validating the efficacy of the LNP delivery system in vivo (Figure 2E).

miRTEN Monotherapy and its Combination with TMZ Inhibits Tumor Growth in Orthotopic GBM

To assess miRTEN’s therapeutic effect in vivo, we used the aggressive orthotopic GBM8 xenograft model. Luc⁺/ mCherry⁺ GBM8 cells were implanted into the striatum of nude mice, and tumor growth was monitored through in vivo bioluminescence imaging. miRTEN (10 µg) was delivered by stereotaxic ICV injection on day 18 (during exponential tumor growth), with a second dose on day 25 (Figure 3A). Control LNPs containing Cas9 mRNA alone or Cas9 + GFP sgRNA showed no significant effect. In contrast, miRTEN significantly slowed tumor progression, with comparable efficacy of G1 or G3 (mean inhibition: 63.6% for G1, 60.2% for G3; Figure 3B and C). Ki67 staining showed reduced proliferation in the tumor core and invasive margins (Figure 3D). qRT-PCR analysis confirmed reduced miR-10b levels in the tumors at experiment’s end (Figure 3E), and de-repression of miR-10b targets (Figure 3F). To test synergy with standard-of-care (SOC) therapy, we combined miRTEN with TMZ. In vitro, this combination significantly enhanced cell death in both GBM8 and GBM6 GSCs, which are relatively sensitive and resistant to TMZ, respectively (Figure 4A and B). In vivo, GBM8-bearing mice received miRTEN (10 µg on day 14, 20 µg on day 21; Figure 4C). Editing efficiency in tumor biopsies collected 5 days later ranged from 40%-75% (G1) and 30%-70% (G3), with minimal regional variability (Figure 4D and E). Indels clustered, as anticipated, in mature miR-10b and Drosha-binding pri-miR-10b regions for G1 and G3, respectively (Supplementary Figure 7A). Editing efficiency strongly correlated with miR-10b suppression (R² = 0.63, P = 0.0004 < 0.001; Figure 4F), which in turn, correlated with tumor growth inhibition (R² = 0.86, P < 0.0001; Figure 4G; Supplementary Figure 7B). In contrast, miR-10b editing in normal brain tissues was low (3%-7% for G1, 2%-6% for G3; Figure 4E).

Figure 3.

LNP-mediated miR-10b GE reduces growth of established intracranial GBM xenografts. (A) FLuc⁺/mCherry⁺ human GBM8 cells (1 × 10⁵) were stereotactically implanted into nude mice brains, and tumor growth was monitored over time. After tumors were established, mice received bilaterally ICV injections of miRTEN (G1 and G3), or control LNPs with Cas9 mRNA and GFP sgRNA on days 18 and 25 post-implantation (10 µg per mouse; n = 9-10 mice per group). (B) Representative luminescence images are shown. (C) Tumor growth, measured as average luminescence per group, is plotted over time. Statistical significance was determined using two-way ANOVA. (D) Representative images of tumors’ Ki67 staining are shown, with quantification of Ki67-positive cells across groups. Statistical significance was determined using one-way ANOVA. (E) miR-10b inhibition in resected tumors was quantified by qRT-PCR, with miR-10b expression normalized to miR-125b. Statistical significance was determined using one-way ANOVA. (F) Expression levels of miR-10b target genes were assessed by qRT-PCR (n = 9-10 mice per group), with mRNA normalized to GAPDH and U6 snRNA to miR-125a. All data are represented as mean ± SEM. *P < 0.05; **P < 0.01; ****P < 0.0001.

Figure 4.

miRTEN synergizes with TMZ to suppress GBM growth in vitro and in vivo. (A) GBM8 and GBM6 cells were treated with increasing concentrations of TMZ (0.78-800 μM) for 5 days. Cell viability was assessed by CellTiter-Glo, and IC₅₀ values were calculated using a 4-parameter logistic regression. (B) Cells were treated with 200 ng miRTEN (G1 or G3) or control LNPs (GFP), with or without 50 μM TMZ, for 5 days. Cell death was measured by 7-AAD flow cytometry. Representative plots and quantification of 7-AAD⁺ cells are shown (n = 3). P-values were determined by one-way ANOVA. (C) FLuc⁺/mCherry⁺ GBM8 cells (1 × 10⁵) were implanted into nude mice brains, and tumor growth was monitored over time. Once tumors were established, mice received bilaterally ICV injections of miRTEN (G1 and G3) or control LNPs with Cas9 mRNA alone on days 14 (10 µg per mouse) and 21 (20 µg per mouse) post-implantations. TMZ (50 mg/kg) were administered on indicated days (n = 10 mice per group). (D) miR-10b GE efficiency was assessed in paired tumor biopsies collected at varying distances from the LNP injection site 5 days after injection (n = 5 mice per group). (E) miR-10b GE efficiency was evaluated in tumors (n = 5 mice per group) and adjacent brain tissue (n = 3 mice per group) 5 days after LNP injection. Statistical significance was determined using one-way ANOVA. (F) The correlation of miR-10b expression, measured by qRT-PCR, and GE, monitored by DNA sequencing, in matching samples (n = 5 mice per group). (G) The correlation between miR-10b expression and tumor growth inhibition was assessed on day 26 using the formula: TGI (%) = (Mean value of relative photon flux in control group − relative photon flux value in individual mice)/Mean value of relative photon flux in control group × 100. (H) Tumor growth was monitored by luciferase bioimaging and plotted for each group. Statistical significance was determined using two-way ANOVA. (I) Kaplan–Meier survival curves are shown for each treatment group. Statistical significance of differences in survival distributions was evaluated with the log-rank (Mantel–Cox) test. All data are represented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significance.

For combination therapy, TMZ (50 mg/kg) was given orally every other day between miRTEN doses (Figure 4H). miRTEN alone inhibited tumor growth, and its combination with TMZ led to further suppression and significantly extended survival. Long-term survival was achieved in 30% (G1) and 40% (G3) of treated mice (Figure 4I), supporting the therapeutic potential of miRTEN, particularly in combination with TMZ, to overcome resistance and improve GBM outcomes.

miRTEN Suppresses GBM Growth and Enhances Antitumor Immune Response in Immunocompetent Mice

Beyond its cell-autonomous effects in glioma,⁵ miR-10b may also modulate the GBM immune response and promote immune evasion.²⁴ To explore this, we analyzed TCGA-GBM dataset (198 samples) with matched mRNA and miRNA profiles. Tumors were classified as immune “hot” (high infiltration), “cold” (low/absent infiltration), or “infiltrating” (moderate infiltration) based on immune cell scores. miR-10b expression was significantly higher in “cold” tumors (Figure 5A and B) and inversely correlated with immune cell infiltration, including CD8⁺ T cells (Figure 5C). Similar trends were seen in the CGGA dataset (Supplementary Figure 8A).

Figure 5.

miR-10b GE inhibits GBM growth and activates antitumor immunity in immunocompetent mice. (A) Data from TCGA-GBM 2008 dataset (cBioPortal, including mRNA expression for 206 samples) was analyzed for immune scores derived from immune-related expression signature.²⁵ Samples were clustered into 3 immune cell score groups, shown in a heatmap. (B) miR-10b expression in 198 GBM samples with matching mRNA and miRNA datasets across 3 groups was analyzed using a rank sum test. (C) Pearson correlation between miR-10b levels and immune cell activity scores. (D) FLuc⁺/mCherry⁺ CT2A mouse glioma cells (5 × 10⁴) were stereotactically implanted in the brains of C57BL/6 mice. Once tumors were established, mice were treated with bilateral ICV injections of miRTEN or Cas9-only control (10 µg on day 8 and 20 µg on day 15, n = 8/group). (E) miR-10b GE efficiency was assessed in tumor samples collected 7-10 days after miRTEN injection (n = 3 mice per group). (F) Representative luminescence images are shown. (G) Tumor growth was monitored by luciferase bioimaging and plotted for each group. Statistical significance was determined using two-way ANOVA. (H) Kaplan–Meier survival curves are shown for each treatment group. (I) IHC of CT2A tumors stained for CD45, CD3, CD4, and CD8 markers shows increased immune cell infiltration in miRTEN-treated tumors (n = 5 samples). Scale bar = 100 µm. (J) qRT-PCR analysis of immunosuppressive and T cell activation markers in resected miRTEN-treated tumors, normalized to GAPDH (n = 8/group). Statistical significance was determined using one-way ANOVA. (K) Schematic of the workflow: naïve CD8⁺ T cells were isolated, subjected to miR-10b GE, polarized to cytotoxic T cells, and analyzed by flow cytometry. (L) Intracellular TNF-α, IFN-γ, and Granzyme B levels were measured by flow cytometry in CD8⁺ T cells post–miR-10b GE (n = 7 mice/group). Statistical significance was determined using one-way ANOVA with Dunnett’s multiple comparison. (M) Kaplan–Meier survival analysis shows that miRTEN-treated animals which recovered from GBM (from G1 and G3 groups in panel H), do not develop tumors after orthotopic re-injection with 5 × 10⁴ CT2A cells. In contrast, matched naïve C57BL/6 mice injected with CT2A cells all succumbed by day 30 (n = 3 per group). All data are represented as mean ± SEM. Statistical significance of differences in survival distributions was evaluated with the log-rank (Mantel–Cox) test. *P < 0.05; **P < 0.01; ***P < 0.001.

We hypothesized that reducing miR-10b levels in GBM might trigger antitumor immune responses and assessed the effects of miRTEN on tumor growth and immune activation in an immunocompetent GBM model. Murine Luc⁺ CT2A glioma cells, known for aggressive growth, were implanted into the brains of C57BL/6 mice, and tumor progression was monitored by the bioluminescence imaging. On day 8 post-implantation, when the tumors were established, mice received 10 µg of miRTEN via stereotaxic ICV injection, followed by 20 µg on day 15 (Figure 5D). Targeted sequencing confirmed efficient miR-10b editing in murine tumors, consistent with strong locus conservation between mice and humans (Supplementary Figure 8B), with editing rates of 30%-55% for G1 and 35%-45% for G3 (Figure 5E). miRTEN treatment markedly slowed tumor growth, with G1 and G3 showing similar efficacy (Figure 5F and G), and extended survival, with 37.5% of mice (3 out of 8) achieving long-term survival (Figure 5H).

Immune profiling showed increased infiltration of total immune cells (CD45⁺) and T cells (CD4⁺ and CD8⁺) in miRTEN-treated tumors (Figure 5I), while NK cells (CD56⁺) and macrophages (CD11⁺) remained unchanged (Supplementary Figure 8C). miRTEN also upregulated T cell activation markers (TNFα, IFNγ, GZMb) and downregulated immunosuppressive factors (IL4, IL10, TGFβ, Tim3, PDL1) (Figure 5J).

To determine whether miR-10b editing has direct effects on CD8⁺ T cells, we isolated naïve CD8⁺ T cells from mouse spleens, treated them with miRTEN, as described (Figure 5K). Edited CD8⁺ cultures exhibited higher frequencies of TNF-α⁺, IFN-γ⁺, and Granzyme B⁺ cells (Figure 5L), along with increased double-positive populations (TNFα⁺IFNγ⁺, TNFα⁺GzmB⁺, IFNγ⁺GzmB⁺) (Supplementary Figure 8D), supporting the role of miR-10b in suppressing cytotoxic T cell activation.

Finally, to test for long-term immune protection, we performed a tumor rechallenge in 6 miRTEN-cured mice. Treatment-naïve mice allografted with CT2A cells developed progressively growing tumors, whereas all miRTEN-treated mice rejected tumor rechallenge (Figure 5M), indicating durable antitumor immune memory.

LNP-mediated miR-10b Gene Editing is Safe and Well-tolerated

Our data demonstrate that LNP formulations selectively enrich Cas9 in GBM tissues while sparing normal brain parenchyma. Prior studies indicate that even when Cas9 is highly expressed, it does not efficiently edit miR-10b in normal neuroglial cells due to the heterochromatic state of the miR-10b locus in nonexpressing cells.¹² To evaluate miRTEN safety, we tested the formulations on primary rat cortical neurons, mixed neuroglial cultures (Supplementary Figure 9A to D), and human iPSC-derived neurons (Figure 6A and B). After miRTEN treatment, only 25%-30% of neurons showed low Cas9 expression (Figure 6A; Supplementary Figure 8A), markedly lower than in glioma and GSC cultures (Figure 1). Even at high doses (200 ng Cas9 mRNA and 200 ng miR-10b G1 or G3 sgRNAs), miRTEN did not impair neuronal morphology or viability. Immunohistochemistry (IHC) for neuronal and astrocyte markers showed no differences between miRTEN-treated, Cas9/GFP sgRNA control, and untreated cells (Figure 6A; Supplementary Figure 9B), and cell viability assays showed no cytotoxicity over time (Supplementary Figure 9C). Neurite length and branching were unaffected in both rat and human neuronal cultures (Figure 6B; Supplementary Figure 9D).

Figure 6.

Toxicity assessment of miR-10b GE therapy in cultured normal neural cells and nontumor-bearing mice. (A) Human iPSC-derived neurons were treated with miRTEN (200 ng) or Cas9/GFP sgRNA-containing LNPs for 5 days (n = 4). MAP2 and Cas9 expression were analyzed by immunostaining. (B) Human iPSC-derived neuron cultures (n = 6) were treated with LNP formulations, monitored by live-cell imaging, and neurite length and branch points were quantified. The data were analyzed using the NeuroTrack module of the Incucyte software. (C) Schematic of the experimental to assess the impact of chronic miRTEN exposure on neuronal activity (top). Human control iPSC-derived neurons (n = 5 wells/ group) were matured via NGN2 induction and transduced with synapsin-driven GCaMP6s. Baseline calcium imaging was performed at DIV31, followed by weekly LNP treatments (200 ng Cas9 mRNA + 200 ng sgRNA per well) and calcium live-cell imaging at DIV39, DIV46, DIV55, and DIV61. (D) Quantification of peak amplitude (40% ΔF/F₀) and inter-spike interval of spontaneous calcium transients. Data analyzed by one-way ANOVA with Sidak’s multiple comparisons. (E) Six-week-old C57BL/6 mice (n = 4/group; 2 males, 2 females) without tumors received bilateral ICV injections of miRTEN (G1 or G3) or LNPs with Cas9 mRNA alone. Doses of 20 or 30 µg of LNPs were administered, and body weight was monitored over time. (F) After 35 days, brain tissues were collected and examined by H&E staining showing no significant histopathological changes in either Cas9-only or miRTEN-treated mice (n = 4/group). (G) Immunofluorescence staining for neuronal (MAP2, NeuN), astrocytic (GFAP), myelin (MBP), and microglial (Iba1) markers was performed. All data are represented as mean ± SEM. ns, no significance.

We further assessed neurotoxicity in human iPSC-derived neurons by tracking neuronal activity using GCaMP6s calcium biosensor live-cell imaging (Figure 6C), a proxy for neuronal activity. A progressive increase in spontaneous calcium uptake from DIV31 to DIV61 confirmed maturation and sustained activity. Four weekly miRTEN treatments had no impact on calcium uptake amplitude, interspike interval (ISI), or neuronal viability compared to Cas9/GFP sgRNA or untreated controls, indicating no adverse effect on neuronal function or survival, even under chronic exposure (Figure 6D; Supplementary Figure 9E).

For in vivo safety assessment, miRTEN (20 or 30 µg) was administered via ICV injections to healthy nontumor-bearing immunocompetent mice (n = 2 male, 2 female), and monitored for 35 days post-injection. LNP treatment caused no acute toxicity, as assessed by quantitative behavioral score (Supplementary Figure 9F; Supplementary Table 3). Mice receiving 30 µg showed a transient 8%-12% body weight loss, recovering by day 7, while the 20-µg group had no weight change, suggesting a dose-dependent effect (Figure 6E).

Editing efficiency in brain tissues was low (1.0%-2.5% for G1; 0.4%-0.75% for G3) (Supplementary Figure 9G). H&E staining of whole brain sections revealed no histopathological abnormalities (Figure 6F). IHC for neuronal, astrocytic, myelin, and microglial markers showed no differences between LNP-treated and PBS control mice (Figure 6G; Supplementary Figure 9H). No abnormal morphology was observed in major organs, including heart, kidney, lung, spleen, and liver (Supplementary Figure 9I). Together, these findings indicate that locally delivered miRTEN is safe, well-tolerated, and causes no off-target toxicity in healthy tissues, underscoring its therapeutic potential for GBM treatment. The absence of tissue responses even at higher doses supports the potential of this LNP-based delivery system for repeated administration in clinical applications.

Discussion

GBM is a highly aggressive malignancy marked by profound intratumoral and interpatient heterogeneity driven by complex genetic and epigenetic alterations.¹ Its immunosuppressive tumor microenvironment (TME)—comprising immune cells, stromal elements, and hypoxia—further complicates treatment.²⁶ Gene therapies, including GE, are emerging as next-generation strategies to target gliomagenesis drivers, but delivery remains a major hurdle. To address this, we developed LNP formulations for efficient RNA delivery and CRISPR–Cas9-based GE in orthotopic GBM, targeting diverse glioma subtypes, including therapy-resistant GSCs, in aggressive human xenografts and mouse allografts models. These formulations were tested for editing a key regulatory target, miR-10b.

MiR-10b is an indispensable regulator of glioma growth across tumor subtypes or mutation statuses, and a highly selective therapeutic target for malignant gliomas.^5-7,9,10 Its consistent expression in GBM and other high-grade gliomas, including GSCs, and absence in normal brain tissue, enable tumor-specific targeting with minimal off-target effects.¹² Inhibiting miR-10b impairs GBM growth and resistance to SOC via 3 mechanisms: tumor-cell-autonomous de-repression of growth-suppressive targets,^5-7 a bystander effect enhancing growth inhibition,²³ and a less-explored activation of antitumor immune responses.^27,28 These findings position miR-10b as a uniquely promising target for GBM therapy.

LNPs have emerged as versatile and highly effective platforms for RNA delivery, exemplified by Onpattro, the first FDA-approved RNAi therapeutic for hereditary transthyretin amyloidosis,²⁹ and the LNP-mRNA COVID-19 vaccines,³⁰ which revolutionized RNA therapeutics. They are also advancing Cas9 mRNA-based genome editing, as seen with Intellia Therapeutics’ NTLA-2001.³¹ These innovations highlight the potential of LNPs for RNA delivery, and their combination with CRISPR has shown initial promise for in vivo GE in preclinical models of cancers such as melanoma,³² liver,³³ and ovarian cancer,³⁴ and for transforming T cells ex vivo for immunotherapy—already in clinical trials.^35,36 However, early reports of CRISPR–Cas9 LNP formulations for GBM therapies were limited by low editing efficiency, insufficient orthotopic model validation, or a lack of demonstrated target engagement.^34,37,38

Ionizable lipids are critical for RNA delivery by influencing LNP encapsulation, cellular uptake, and endosomal escape.³⁹ Clinically approved lipids like DLin-MC3-DMA (MC3) (Patisiran) and SM-102 (Moderna’s vaccines) set the standard, but LP01 surpasses MC3 in nucleic acid encapsulation efficiency⁴⁰ and exhibits significantly lower tissue toxicity due to superior biodegradability, preventing cytotoxic accumulation.^18,41,42 This makes LP01 a strong candidate for clinical GE in vivo. Finn et al. demonstrated that LP01-based LNPs with optimized helper lipids (DSPC, DOPE), cholesterol, lipid-PEG, efficiently edit liver gene, leading to LNP-INT01 development for ATTR amyloidosis.^18,31

We compared MC3, SM-102, and LP01 and found that LP01-based LNPs effectively and selectively target GSCs in vitro and in vivo, showing broad Cas9 distribution mainly in GBM tissues with minimal brain parenchyma expression, enhancing precision and reducing off-target risks. Tumor-specific accumulation is attributed to enhanced permeability and retention effects from leaky vasculature and poor lymphatic drainage.^43,44 Additionally, cationic LNPs facilitate electrostatic interactions with negatively charged tumor-cell membranes, enhancing cellular uptake and GE efficiency.^45,46

LNP-mediated GE showed remarkable durability in vivo, lasting up to 12 months in liver^18,47 and 660 days in lung epithelial cells,⁴⁸ though durability in cancer tissues remained unknown. Here, we show Cas9 expression in orthotopic GBM for at least 3 weeks despite the tumor aggressive growth, likely underestimating persistence in less proliferative human tumors. Two miR-10b sgRNAs—G1 targeting mature miRNA and G3 targeting pri-miRNA—achieved up to 75% editing efficiency, reducing miR-10b levels by 85%, significantly inhibiting tumor growth, and clearing tumors in 37.5% of immunocompetent mice. Notably, strong correlation between editing and miR-10b reduction confirms target engagement and supports miR-10b levels as a quantitative biomarker for miRTEN therapies.⁴⁹ Even the lowest editing efficiencies exceeded the 20% threshold needed to trigger a significant bystander effect.²³

Nevertheless, optimization of miRTEN dose and administration remains critical for clinical translation. mRNA therapies often require repeated administrations due to limited stability.⁵⁰ We used Cas9 mRNA stabilized with 5moU, while additional modifications like N1-methyl pseudo-U could further enhance stability and translation.⁵¹ Clinical protocols for Onpattro and COVID-19 vaccines involve multiple administrations over weeks to months,^52,53 and preclinical mRNA/LNP studies support repeated dosing intervals from days to weeks without systemic toxicity.^54-57 Our findings indicate a dose-dependent effect, with 2 ICV injections of 10 and 20 µg effectively suppressing miR-10b and tumor growth. Importantly, both in vitro and in vivo analyses confirmed the safety of miRTEN, as repeated treatments did not impair neuronal activity in human neurons nor cause detectable neurotoxicity in mouse brain, supporting its suitability for CNS-targeted gene editing. Targeting GSCs—the most therapy-resistant cells driving recurrence—may effectively prevent tumor regrowth. Future investigations will refine dosing to maximize efficacy and safety.

Beyond its oncogenic role, miR-10b has been implicated in T cell^– and macrophage-mediated immune responses across diseases.^58-61 In other cancers, it promotes immune evasion by suppressing NK cell activation, regulating T cells, and targeting genes like p21, p16, and Bim, thereby fostering an immunosuppressive TME.^27,62-64 It has also been proposed as a TET2-mediated regulator of PD-L1 in GBM.²⁴ We show that miRTEN treatment increased immune cell infiltration, particularly cytotoxic CD8⁺ T cells, and established durable immune memory in the brain. High miR-10b levels correlate with immune-resistant “cold” GBM tumors marked by low immune cell infiltration and minimal antitumor activity.⁶⁵ Editing miR-10b upregulated T cell activation markers (TNFα, IFNγ, GZMb) and downregulated immunosuppressive factors (PD-L1, IL4, IL10, TGFβ, Tim3), effectively converting “cold” to “hot” tumors—a major immunotherapy goal.⁶⁶ Thus, miRTEN not only suppresses tumors directly but also appears to limit tumor growth by boosting antitumor immunity. The data showing miR-10b is expressed in T cells,^58,60,67 and the enhanced cytotoxicity of miR-10b-edited CD8⁺ T cells suggest that miRTEN’s impact is partly mediated by direct regulation of T cell function. Based on this observation, we postulate that a combination of miRTEN with immune checkpoint inhibitors may help overcome GBM immune resistance and enhance therapeutic efficacy.

We also tested miRTEN in combination with TMZ, the standard GBM chemotherapy. Recurrent GBM often resists TMZ, but miR-10b inhibition has been shown to sensitize glioma cells to the drug.⁶⁸ Our experiments confirmed that the combination significantly enhanced tumor suppression in orthotopic models, extending survival and achieving cures in ~45% of GBM8-bearing mice. This suggests miRTEN may benefit both newly diagnosed and recurrent GBM patients by enhancing TMZ sensitivity.

Overall, this study highlights miRTEN’s potential to tackle GBM by addressing tumor heterogeneity, immune evasion, and drug resistance. While only a few in vivo CRISPR-based RNA-LNP therapies are in clinical trials (e.g., for inherited blindness), substantial characterization and analytics are still required for clinical translation. However, the developed drug formulation is safe and scalable, leveraging LNP technologies already validated in mass production and clinical applications.^52,53 Given miR-10b’s unique properties as a target, miRTEN could advance as one of the first in vivo GE therapies for malignant gliomas. More broadly, it offers a platform for RNA-based therapies, presenting a promising strategy for GBM and other challenging malignancies.

Supplementary material

Supplementary material is available online at Neuro-Oncology (https://academic.oup.com/neuro-oncology).

Funding

This work was supported by the R01 CA215072, National Brain Tumor Society, and StacheStrong grants to A.M.K.

Conflicts of interest. All authors declare no competing interests.

Author contributions

A.M.K. conceived and designed the study; Y.Z. performed experiments, data analysis, and visualization; Ami K assisted with primary neuron cultures; Anastasia K. and V.K. assisted with human neuron preparation; D.R. and J.A.K. assisted with neuronal activity assessment; J.F.V. and G.M. assisted with immune cell analysis; D.M. and R.R. assisted with LNP formulations; R.R., E.D., E.J.U. contributed to data analysis; Y.Z. and A.M.K. wrote the manuscript. All authors reviewed and approved the manuscript. A.M.K. acquired funding and supervised the study.

Ethics statement

All animal studies were carried out in accordance with guidelines by the Institutional Animal Care and Use Committee at Brigham and Women’s Hospital.

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.