Granulocyte-Macrophage Colony-Stimulating Factor mRNA and Neuroprotective Immunity in Parkinson’s Disease

Abstract

Restoring numbers and function of regulatory T cells (Tregs) is a novel therapeutic strategy for neurodegenerative disorders. Whether Treg function is boosted by adoptive cell transfer, pharmaceuticals, or immune modulators, the final result is a robust anti-inflammatory and neuronal sparing response. Herein, a newly developed lipid nanoparticle (LNP) containing mRNA encoding granulocyte-macrophage colony-stimulating factor (Gm-csf mRNA) was developed to peripherally induce Tregs, used for treatment in preclinical Parkinson’s disease (PD) models. Administration of Gm-csf mRNA to 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-treated mice and rats overexpressing alpha-synuclein produced dose-dependent increases in plasma GM-CSF levels and peripheral CD4+CD25+FoxP3+ Treg populations. This upregulation paralleled nigrostriatal neuroprotection, upregulated immunosuppression-associated mRNAs that led to the detection of a treatment-induced CD4+ T cell population, and decreased reactive microgliosis. The current findings strengthen prior works utilizing immune modulation by harnessing Gm-csf mRNA to augment adaptive immune function by employing a new delivery platform to treat PD and potentially other neurodegenerative disorders.

Introduction

Phenotypic shift of T cell subsets is being developed for treatment of Parkinson’s disease (PD) and other neurodegenerative disorders [1–7]. For PD, disease onset and progression are linked to reduced Treg numbers and anti-inflammatory function [8]. Whether pharmaceutical interventions serve to correct disease-linked immune deficits remains under active investigation [1, 9–13]. An overarching question remaining is how progressive nigrostriatal dopaminergic neurodegeneration is affected by dysfunctional immunity. Neuroinflammation and aberrant adaptive immune responses are linked to changes in the brain microenvironment leading to disease onset and progression [11, 14–16]. Potent neuroinflammatory activities in PD are reflected through the presence of HLA-DR+ reactive microglia within the substantia nigra (SN), decreased mitochondrial function, increased proinflammatory IL1β, IL2, IL6, TNFα, IFN□, NOS, and ROS within the striatum, SN, peripheral blood and cerebral spinal fluids (CSF) [17–19]. Both CD4+ and CD8+ T cells are detected in the proximity of dopaminergic neurons. This occurs in parallel with decreased levels of peripheral naïve T and effector lymphocytes, respectively. This along with decreased Treg function support a disease-associated adaptive immune dysregulation [19–21]. Moreover, the blood-brain barrier (BBB) is altered as a consequence of neuroinflammation, affecting its leakiness and accelerating immunocyte (monocyte-macrophage and lymphocyte) penetrance. [16, 22]. Any transformation of an aberrant peripheral immune response could facilitate disease interdiction.

Recently, we, and others, have demonstrated that decreasing the brain’s inflammatory microenvironment through the induction of Treg can slow and/or attenuate neuronal loss [4, 7, 23, 24]. Specifically, CD4+ Tregs modulate neuroinflammation and lead to neuroprotective responses, whereas CD4+ Teffs increase the tempo of disease [4, 5]. In PD, diminished Treg numbers and function parallel disease severity [8]. In contrast, increasing Treg function and/or number leads to neuroprotective disease outcomes that translate into improvements in Unified Parkinson’s Disease Rating Scale (UPDRS) part III scores and clinical outcomes [4–8]. Thus, Treg induction has been promoted as a therapeutic strategy resting in its abilities to control neuroinflammation and prevent dopaminergic neuronal loss in PD [9, 24].

The search for pharmacological inducers of Treg led to the novel use of granulocyte-macrophage colony-stimulating factor (GM-CSF) [6, 10]. Therein, we showed that GM-CSF increased Treg numbers and its immunoregulatory activities that led directly to dopaminergic neuronal protection in the substantia nigra and striatum [6]. The translational potential of GM-CSF (Leukine, sargramostim) was recently affirmed in a Phase I PD clinical trial [10]. GM-CSF administered for 2 months led to increased Treg numbers and function, and improved clinical motor activity. However, while daily treatment was generally well-tolerated, moderate adverse events were noted including injection site reactions, elevated white blood cell counts, and muscle and bone pain [10, 25–27]. Likewise, the limited bioavailability and short half-life of GM-CSF required high and frequent doses [27–29]. Thus, alternative strategies in formulations and GM-CSF delivery are required. Therefore, a unique lipid nanoparticle (LNP) containing Csf2 (Gm-csf mRNA) was developed as an alternative to direct administration of GM-CSF protein, providing the potential to decrease the amount and frequency of dose, while generating a similar neuroprotective profile. Here, we demonstrate an alternative drug delivery system leading to increased Treg numbers, cellular function, and potent neuroprotective activities in PD models of human disease following mRNA treatments. Cytokine profiling and transcriptomic analysis by single-cell RNA sequencing was utilized to confirm shifts in innate and adaptive immune populations following treatment. These findings warrant additional steps toward clinical translation of immune modulating mRNA treatment and have broad potential for a range of neurodegenerative diseases.

Materials and Methods

mRNA synthesis and formulation

Gm-csf mRNA, encoding either the Mus musculus or Rattus norvegicus precursor protein, was synthesized by in vitro transcription (IVT). A plasmid encoding the T7 RNA polymerase promotor followed by 5’ untranslated region (UTR), open reading frame (ORF), 3’UTR, and polyA tail was overexpressed in E. coli, linearized, and purified to homogeneity. The mRNA was synthesized by IVT using T7 RNA polymerase, where uridine triphosphate was substituted with N1-methyl-pseudouridine triphosphate [30]. Cap 1 was utilized to improve translation efficiency. After the transcription reaction, mRNAs were purified, buffer exchanged into sodium citrate buffer, and stored at −20 °C until use. Control mRNA NTFIX (nontranslated Factor IX) was synthesized by parallel methods. Lipid nanoparticle (LNP) formulations were prepared by modifying a described method [31]. Formulation of mRNA was performed through ethanol injection nanoprecipitation by mixing acidified RNA and ionizable, fusogenic, structural, and PEG lipids dissolved in ethanol at a 3:1 ratio (aqueous:ethanol). After pH adjustment, the mRNA-loaded lipid nanoparticles were buffer exchanged into PBS (pH 7.4), transferred to vials, and stored refrigerated at 5 °C or frozen at −20 °C until further use. A release panel of analytical characterization assays were performed, which included particle size and polydispersity, encapsulation, and endotoxin content before the material was deemed acceptable for in vivo study. mRNA and LNP used in this study exhibit >6 months of stability under frozen conditions and are retested bi-annually to extended shelf-life. Final particle size and encapsulation were <100 nm and >80%, respectively, with endotoxin below 10 EU/mL.

Animals, mRNA treatment, MPTP intoxication, and alpha-synuclein overexpression

Animals were housed, maintained and used for experiments following guidelines set forth by the National Institutes of Health Institutional Review Board and approved by the Animal Care and Use Committee of the University of Nebraska Medical Center. For mouse studies, C57BL/6 mice (6–8 weeks old) were obtained from Jackson Laboratories (stock # 000664). After acclimation, mice were injected intramuscularly with a LNP containing Mus musculus Gm-csf mRNA (Moderna, Inc., Cambridge, MA) (Gm-csf mRNA). For dose response studies, mice were injected daily for 4 consecutive days at doses ranging from 0.00001 mg/kg to 0.1 mg/kg LNP-Gm-csf mRNA or 0.1 mg/kg LNP-NTFIX. For neuroprotection experiments, mice were injected with either vehicle (DPBS, 10 ml/kg body weight) or MPTP reconstituted in phosphate buffered saline (PBS) obtained from Sigma-Aldrich, St. Louis, MO. Mice received 4 subcutaneous injections of MPTP-HCl (16 mg free base/kg), each administered at 2-hour intervals. MPTP safety precautions were followed in accordance with determined safety and handling protocol [32]. On days two and seven after MPTP intoxication, mice were sacrificed and brains were harvested for neuroinflammation evaluations (day 2) and neuronal survival (day 7), respectively (Supplemental Figure 1). For rat studies, 7-week old male Sprague-Dawley (SASCO) rats were ordered from Charles River Laboratories. Rats were injected intramuscularly with lipid nanoparticles containing Rattus norvegicus Gm-csf mRNA (Gm-csf mRNA) (Moderna, Inc., Cambridge, MA). For naïve rat studies, animals were injected for 4 consecutive days with either 0.01 mg/kg or 0.1 mg/kg Gm-csf mRNA and sacrificed on day 5. For human alpha-synuclein (α-Syn) overexpression studies, rats were injected for 4 consecutive days immediately following stereotactic injection, followed by injections every other day until sacrifice. On day 28, rats were sacrificed, and spleens and brains were harvested (Supplemental Figure 1).

Stereotactic Injection

Sprague-Dawley rats were anesthetized with 2% isoflurane in O2 and placed in a stereotaxic device (Leica Biosystems Inc., Buffalo Grove, IL) to secure their skulls. Following skull exposure and formation of a 1–2 mm hole, a sterile Hamilton syringe (model 8100, Thermo Fisher) attached to a 26-gauge needle was inserted into the brain. Vectors were delivered via syringe pump. For α-Syn overexpression, AAV2/1-CBA-HuαSyn-IRES-eGFP-WPRE (Standaert-5713) vector (AAV-α-Syn) and control AAV2/1-IRES-eGFP-WPRE (Standaert-5712) vector (AAV-GFP) were obtained from the University of Iowa (Vector Core, Iowa City, IA) with kind permission from Dr. David G. Standaert (Department of Neurology, University of Alabama-Birmingham, Birmingham, AL) [33, 34]. In 3 □ of PBS, 3×10⁹ genomic copies of AAV-vectors were delivered to the left hemisphere above the substantia nigra at the following coordinates relative to the bregma: AP,−5.3 mm; ML, −2.0 mm; DV, −7.5mm DV.

Perfusions and immunohistochemistry

Under terminal anesthesia (Fatal Plus, pentobarbital), mice and rats were perfused via cardiac puncture with DPBS followed by 4% paraformaldehyde (PFA) (Sigma-Aldrich) in DPBS. Whole brains from mice 7 days post MPTP or from rats 28 days post AAV stereotactic injection were harvested after perfusion to assess survival of dopaminergic neuron cell bodies in the substantia nigra (SN) and termini in the striatum. Frozen midbrains were sectioned at 30 μm and were immunostained for tyrosine hydroxylase (TH) (anti-TH, 1:2000, EMD Millipore) and counterstained for Nissl substance [35]. To assess microglial reactivity, brains were harvested 2 days after MPTP (mice) or 28 days after stereotactic injection of AAV (rats) and midbrain sections (30 μm) were immunostained for Mac-1 (anti-CD11b, 1:1000, AbD Serotech) for mice and Iba-1 (1:1000, VWR) for rat. To assess dopaminergic termini, striatal sections (30 μm) were labeled with anti-TH (1:1000, EMD Millipore). To visualize antibody-labeled tissues, sections were incubated in streptavidin-HRP solution (ABC Elite Vector Kit, Vector Laboratories) and color was developed using an H₂O₂ generation system in the presence of diaminobenzidine (DAB) chromogen (Sigma-Aldrich). Estimated neuron and reactive microglial numbers were quantified by a blinded investigator and unbiased stereological analysis using StereoInvestigator software (MBF Bioscience) [35]. Density of dopaminergic neuron termini in the striatum was determined by digital densitometry using Image J software (National Institutes of Health).

Regulatory T Cell Suppression Assays

For mouse studies, CD4+CD25+ and CD4+CD25- cells were isolated from spleen using EasySep Mouse CD4+CD25+ Regulatory T Cell Isolation Kit II (StemCell) per the manufacturer’s instructions. For rat studies, the same cell populations were isolated using EasySep Rat CD4+ T Cell Isolation Kit (StemCell). Isolated rat CD4+ cells were stained with anti-CD25 PE (BD Bioscience) for 20 minutes at a concentration of 0.75 μg/ml per 1.5 × 10⁷ cells. Anti-PE-magnetic beads from EasySep PE Positive Selection Kit II (StemCell) were then added for the positive magnetic separation of CD4+CD25+ T cells. Isolated cell populations from mouse and rat were assessed for purity by flow cytometric analysis and were determined to be > 90% CD4+CD25+ and >60% CD4+CD25+FOXP3+ as determined by the expression of intracellular FOXP3. The CD4+CD25- T cell fraction was collected from the PBS-treated group (0 mg/kg) in both rats and mice and served as the Tresponder (Tresp) population for the suppression assay. Isolated Tresps were labeled with carboxyfluorescein succinimidyl ester (CFSE) (Thermo Fisher) in order to track cell proliferation. Tregs were serially diluted by 2-fold into wells of a 96-well U bottom microtiter plate and CFSE-stained Tresps were plated at a concentration of 50 × 10⁶ cells/ml to yield Treg:Tresp ratios of 2:1, 1:1, 0.5:1, 0.25:1, and 0.125:1. Mouse cells were stimulated for proliferation using Dynabeads T-activator CD3/CD28 beads (ThermoFisher) at a 1:1 bead:cell ratio. For rat cell stimulation, Dynabeads M-450 epoxy (ThermoFisher) were conjugated using anti-rat CD3 and anti-rat CD28 according to the manufacturer’s protocol. The resulting CD3:CD28 ratio was 1:1 and the resulting bead:antibody ratio was 1000 beads:200 μg (100 μg of each antibody). After conjugation, beads were stored at 4°C at a concentration of 4 × 10⁷ beads/ml in PBS, pH 7.4 with 0.1% bovine serum albumin (BSA). Stimulated Tresps alone and unstimulated Tresps were plated as controls. Suppression assay cultures were incubated at 37°C in 5% CO₂ for 3 days, fixed, and analyzed on a BD LSRII flow cytometer. The extent of proliferation using CFSE fluorescence was assessed utilizing FACSDiva Software (BD Biosciences, San Jose, CA). Treg-mediated inhibition was calculated as % Inhibition = 1 – (% Proliferating Tresp:Treg ÷ % Proliferating Stimulated Tresp Alone).

Adoptive Transfer

From donor mice treated with Gm-csf mRNA, splenic CD4+CD25+ cells were isolated using the same kits as described in the suppression assay. Recipient mice were intoxicated with MPTP as described and 1 × 10⁶ CD4+CD25+ cells were adoptively transferred via tail vein injection 8 to 12 hours post-MPTP treatment. On days 2 or 7 following administration of MPTP, mice were sacrificed, and brains were harvested and processed (Supplemental Figure 1).

Flow Cytometric Assessments

After 4 days of Gm-csf mRNA treatment, whole blood and spleens of rats and mice were collected to determine T cell and B cell profiles via flow cytometric analysis. Whole blood (50 μl) and splenocytes (1 × 10⁶) were fluorescently labeled using antibodies against surface markers for CD3, CD4, CD25, CD8, and the intracellular marker for FOXP3. Mouse blood and splenocytes were labeled with PerCP-Cy5.5-anti-CD3 (eBioscience), PE-Cy7-anti-CD4 (eBioscience), PE-anti-CD25 (eBioscience) and FITC-anti-CD8 (eBioscience). Rat blood was stained with BV-421-anti-CD3 (BD Bioscience), PerCP-eFluor710-anti-CD4 (eBioscience), BV-786-anti-CD8 (BD Bioscience), and PE-anti-CD25 (BD Bioscience). For intracellular staining of both rat and mouse cells, cells were permeabilized for 45 min at 4°C using FOXP3/Transcription Factor Staining Buffer Set (eBioscience). Cells were then labeled with APC-anti-FOXP3 (eBioscience) followed by fixation. Samples were processed on a BD LSRII flow cytometer and analyzed using FACSDiva Software (BD Biosciences, San Jose, CA). Cell frequencies were determined from the total lymphocyte population. For myeloid population immunophenotyping, mouse spleens were collected and fluorescently labeled using the following antibodies: Pacific-blue-anti-CD11b (Biolegend), Alexa-fluor-700-anti-CD11c (Biolegend), PerCP-anti-I-A/I-E (Biolegend), PE-anti-CD86 (Biolegend), PE-Cy7-anti-CD86 (Biolegend), anti-CD-80 (abcam, ab134120) and FITC-anti-CD40 (Biolegend). Data were acquired on a flow cytometer (LSRFortessa™, FACSCantoII, BD Biosciences, San Jose, CA) and analyzed with FlowJo® (FlowJo, LLC, Ashland, Oregon).

Blood Chemistry and Peripheral Blood Assessments

At the time of sacrifice, 250 μl of whole blood was collected into K₂EDTA blood collection tubes for complete blood count (CBC) levels or into heparinized blood collection tubes for blood chemistry and metabolite levels. Following isolation, heparinized blood was centrifuged, and plasma was collected. Complete metabolic panels were carried out using VetScan Chemistry Comprehensive Test cartridges (Abaxis) on a VetScan VS2 machine. For CBC analysis, whole blood collected from K₂EDTA tubes was immediately assayed on a VetScan HM5 machine.

GM-CSF Protein Quantification

Prior to treatment, peripheral blood from mice was collected via maxillary bleed. Mice were then treated with Gm-csf mRNA, and after 6 hours, mice were bled again. Plasma was collected by centrifuging whole blood at 10,600 x g for 10 minutes and stored at −80°C. After 4 days of Gm-csf mRNA treatment, mice were sacrificed and the spleen, liver, brain, and inguinal lymph nodes were harvested. These were flash frozen. Five mg of tissue was lysed with NP40 (Cell Lysis Buffer, Invitrogen) containing an EDTA-free protease inhibitor (Sigma Aldrich). Samples were sonicated, centrifuged then aliquoted. Following collection and storage, a Pierce BCA Protein Assay (ThermoFisher) was performed to determine total protein concentration. GM-CSF protein levels were then determined using Mouse GM-CSF Quantikine ELISA kit (R&D Systems) using the manufacturer’s protocol.

Cytokine Assessments

Before α-Syn overexpression and Gm-csf mRNA treatment, rats were bled via maxillary vein, and peripheral blood was collected as a baseline. After collection, blood was centrifuged at 1000 x g for 10 minutes, and plasma was collected and stored at −80°C. After 28 days, at the time of sacrifice, blood and plasma were collected again. After collection, levels of cytokines within plasma before and after treatment were assessed using Rat Cytokine Array Panel A (R&D Systems) protocol.

Single-cell RNA sequencing and data analysis

Five days post-treatment with either PBS or 0.1 mg/kg Gm-csf mRNA, spleens were harvested and CD4+ cells were enriched by negative selection using the EasySep Mouse CD4+ T Cell Isolation (StemCell) per the manufacturer’s instructions. Enriched samples were then prepared and sequenced in the UNMC Genomics Core Facility. Briefly stated, cell suspensions were evaluated by light microscopy to determine concentration, percent viability, and assess potential debris present in the suspension. Cells were loaded onto a 10x Genomics instrument and single cells were captured, lysed, RNA reverse transcribed, and RNA barcoded using Chromium Single Cell 3’ Reagent Kits v3 reagents following the manufacturers’ suggested protocol (10x Genomics, Pleasanton, CA). Illumina compatible cDNA libraries were created and quantified by qPCR using the KAPA Library Quant Kit (Illumina) from KAPA Biosystems (Roche, Pleasonton, CA). Samples were sequenced to an average depth of approximately 54,000 mean reads per cell on an Illumina NovaSeq6000 instrument. Approximately 7,500 (Gm-csf mRNA) and 7,900 (PBS) cells per sample were subjected to the analysis. FASTQ basecall files were provided to the UNMC Bioinformatics and Systems Biology for further analysis. Cellranger (v.5.0.1) mkfasq was applied to the basecall files to generate sequencing and 10x-specific quality control metrics, including barcode quality, accuracy, and diversity. FASTQ files were generated and inserted into the Cellranger count pipeline to perform alignment using STAR aligner with MM10 reference genome, filtering, and UMI (Unique molecular identifier) counting. Chromium cellular barcodes were used to generate gene-barcode matrices and perform clustering and gene expression analysis. Gene expression values were then normalized to a reference genome, scaled, and log transformed. Clustering methods were determined as K-means using an unsupervised graph-based method. Principal Components Analysis (PCA) was utilized to change the dimensionality of the dataset to a user-selectable number of principal components, followed by t-SNE (t-Stochastic Neighbor Embedding) to visualize the data in two-dimensional space. Cellranger aggregation was performed to normalize and pool the results for treatment-based comparisons. Following intital aggregation and cluster analysis, CD3 and CD4 co-expression levels were assessed on all cell populations. Clusters were then filtered to include only UMIs from CD4+CD3+ subsets due to the mixed starting population, thus allowing for analysis of only CD4+ T cells.

Statistical Analyses

For all studies, data were analyzed using GraphPad Prism 7.0 software (La Jolla, CA). All values are expressed as mean ± SEM. Differences in between-group means were analyzed using one-way ANOVA followed by Tukey or Newman-Keuls post hoc test, depending on assay. Significant difference for all studies was selected at P < 0.05. Measurement of Treg function and dose-dependency were assessed by linear regression analyses as either a function of Treg:Tresp ratio or Gm-csf mRNA dose. Differences in Treg suppressive function were determined by differences between groups in slope or intercept. Slopes for all lines were determined to be significantly non-zero. For single-cell RNA sequencing, cluster analysis was determined using statistical significance of the expression difference and was based on a negative binomial test. P-values were adjusted for multiple testing via the Benjamini-Hochberg procedure.

Results

Gm-csf mRNA induces detectable GM-CSF protein in plasma within six hours after intramuscular injection of 0.01, 0.05, and 0.1 mg/kg Gm-csf mRNA (Figure 1A). Treatment with 0.01 mg/kg Gm-csf mRNA significantly increased plasma GM-CSF levels from 188 pg/ml to 1487 pg/ml, treatment with 0.05 mg/kg significantly increased levels from 73 pg/ml to 6215 pg/ml, and treatment with 0.1 mg/kg significantly increased levels from 12 pg/ml to 8211 pg/ml. However, elevated GM-CSF protein levels were not detected in plasma of mice treated with 0.001 mg/kg or lower doses of Gm-csf mRNA or those treated with a non-translatable mRNA negative control (NTFIX). Variation in endogenous GM-CSF levels were recorded among the groups. Normative serum values for mice generally average 40 pg/ml with ranges of 8–128 pg/ml for mice [36]. However, in this instance, some endogenous readings fell slightly higher than normal. This may reflect assay variability, animal to animal variations or lower doses administered in replicate study experiments. GM-CSF protein was not significantly increased in spleen, liver, brain, or inguinal lymph nodes of mice treated with Gm-csf mRNA (< 71 pg/ml GM-CSF) compared to tissues isolated from untreated animals (10–50 pg/ml GM-CSF) (data not shown). Treatment with ascending doses of Gm-csf mRNA resulted in significant enlargement in spleen size that was found to be dose-dependent by linear regression analysis (R² = 0.46, P = 0.0009) (Figure 1B and C). Along with increased spleen size, parallel increases in peripheral white blood cells (WBC) were observed (Figure 1D-G). Absolute blood cell counts revealed an increasing trend in WBC (Figure 1D), monocytes (Figure 1E) and neutrophils (Figure 1F) that coincided with increasing doses of Gm-csf mRNA but were only significantly increased with 0.1 mg/kg dose. However, lymphocyte populations were only slightly affected (Figure 1G). Overall, comprehensive metabolic panels were not majorly altered by treatment as levels of creatinine, total bilirubin, alanine aminotransferase, glucose, urea nitrogen, sodium, calcium, globulin, and phosphorus were unchanged by treatment, whereas alkaline phosphatase (Figure 1H), albumin (Figure 1I), and amylase (Figure 1J) decreased in 0.01 and/or 0.1 mg/kg-treated animals. NFTIX treatment did not significantly differ from PBS (0 mg/kg) in any assessments.

Figure 1. Intramuscular injection of Gm-csf mRNA results in elevated GM-CSF protein levels, increased spleen size, altered blood counts and chemistry profiles.

A. Quantification of plasma GM-CSF protein levels in peripheral blood before (pre) and 6 hours after (post) initial injection with multiple ascending doses of Gm-csf mRNA construct or NTFIX. B. Representative images depicting spleen enlargement in mice treated with multiple ascending doses of Gm-csf mRNA construct or NTFIX. C. Quantification of organ weight four days after initial treatment, and linear regression analysis of organ weight (R² = 0.4674, P = 0.0009). D-G. Absolute counts of white blood cells (WBC) (D), monocytes (E), neutrophils (F), and lymphocytes (G) within whole blood following treatment. H-J. Significant changes in blood chemistry profiles for alkaline phosphatase (H), albumin (I) and amylase (J) following treatment. All other measured metabolites (creatinine, total bilirubin, alanine aminotransferase, glucose, urea nitrogen, sodium, calcium, globulin, and phosphorus) remained unchanged with treatment. A, C-J. Differences in mean ± SEM (n = 4–5 per group) were determined where P < 0.05 compared with *pre, and ^a0 mg/kg.

Gm-csf mRNA increases Treg and myeloid populations leading to neuroprotection in 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-

Next, T cell phenotypes within peripheral blood were assessed via flow cytometric analysis (Figure 2A). Treatment of mice with 4 consecutive daily doses of LNP-formulated Gm-csf mRNA ranging from 0.001 mg/kg to 0.01 mg/kg showed no significant change in T cell frequencies within peripheral blood (Figure 2B-D). However, 0.1 mg/kg treatment significantly reduced CD3+, CD8+, and CD4+ T cell frequencies from those observed in untreated animals. In contrast, 0.1 mg/kg treatment significantly increased CD4+CD25+FOXP3+ cell frequencies from 3.5 ± 0.49 to 14.3 ± 0.99% compared to untreated controls (Figure 2E). Treatment at 0.001 and 0.01 mg/kg Gm-csf mRNA also increased CD4+CD25+FOXP3+ cell frequencies to 8.2 ± 1.6 and 9.2 ± 2.0 %, but the elevation was not significant. NTFIX treatment had no effect on T cell subsets. Additionally, splenic Tregs isolated from Gm-csf mRNA-treated animals suppressed the proliferation of CD3/CD28-stimulated, carboxyfluorescein succinimidyl ester (CFSE)-labeled CD4+CD25- T responder cells (Tresps) (Figure 2F-H) A representative flow cytometric readout of Treg-induced inhibition of Tresp proliferation demonstrates loss of inhibitory function with dilution of Tregs (Figure 2F). The percent of proliferating Tresp in co-culture with Tregs was determined (Figure 2G) and the resulting Treg inhibitory capacity was calculated (Figure 2G and 2H) [8, 37]. Significant elevations in slope indicate greater inhibitory capacity at higher Tresp:Treg ratios. Determination of elevation using linear regression analysis indicated significantly enhanced suppressive capacity of Tregs isolated from mice treated with Gm-csf mRNA at 0.01 mg/kg (p = 0.007) or 0.1 mg/kg (p = 0.04) compared to CD4+CD25+ Tregs from PBS-treated controls.

Figure 2. Gm-csf mRNA dose escalations elicit increases in CD4+CD25+FOXP3+ Treg numbers.

A. Representative gating strategy for flow cytometric analysis of T cell subsets within the lymphocyte population of whole blood in mice treated with multiple ascending doses of Gm-csf mRNA construct or NTFIX. Quantification of CD3+ (B), CD8+ (C), CD4+ (D) within the lymphocyte population, and CD4+CD25+FOXP3+ (E) cells within CD4+ populations in peripheral blood following i.m. injection with PBS (0 mg/kg), 0.0001 mg/kg, 0.001 mg/kg, 0.01 mg/kg, and 0.1 mg/kg Gm-csf mRNA, or 0.1 mg/kg NTFIX. F. Representative histograms depicting CFSE+ fluorescence at varying Tresp:Treg ratios for determining Tresp proliferation and Treg-mediated inhibition. Tresp were isolated from the spleen of PBS-treated animals only, and Tregs were isolated from each treatment group. CFSE staining intensity is displayed on the x axis. Purple peaks represent the unstained Treg population, and blue peaks represent CFSE-stained parent Tresp populations. Green peaks represent proliferating Tresp populations and are used to calculate % Proliferation. G. Table of mean % Tresp Proliferation ± SEM and mean % Treg inhibition for each treatment dose and Tresp:Treg ratio. H. Assessment of Treg-mediated inhibition (±SEM) of CFSE-stained Tresps (CD4+CD25-) stimulated with anti-CD3/CD28 beads. Tregs were isolated from untreated (0 mg/kg) or Gm-csf mRNA-treated (0.001 mg/kg – 0.1 mg/kg) after 4 days of treatment. Linear regression analysis indicates P < 0.04, R² ≥ 0.91 for 0 mg/kg, 0.001 mg/kg, and 0.01 mg/kg. A-G. Differences in mean ± SEM (n = 4–5 per group) were determined where P < 0.05 compared with ^a0 mg/kg.

In addition to the changes in Treg frequency described above, significantly increased frequency of splenic CD11c⁺MHCII⁺ classical dendritic cells (cDC) were found in mice treated with Gm-csf mRNA dosed at 0.001–0.1 mg/kg (Figure 3A and 3B). However, only the 0.1 mg/kg Gm-csf mRNA treatment group displayed significant expansion of CD11b+Cd11c- myeloid population (Figure 3C). To determine whether LNP-formulated Gm-csf mRNA alters maturation status of the expanded CD11c+ dendritic cell (DC) population, expression of maturation markers such as CD80, CD86, CD40 and Class II (IA-IE) were evaluated by flow cytometry by determining Mean Fluorescence Intensity (MFI). Increased expression of CD86 per cell, as indicated by increased CD86 MFI, was observed in the 0.01 mg/kg Gm-csf mRNA treated group compared to PBS controls; while an increased CD80 MFI is observed in both treatment groups when compared to the PBS control (Figure 3D). MFI of class II was significantly reduced upon treatment with 0.01 or 0.1 mg/kg Gm-csf mRNA.

Figure 3. Gm-csf mRNA increases myeloid numbers in a dose-dependent manner.

A and B. Frequency of CD11c+ and representative flow cytometry plots gated on total CD45+CD3- population in control (LNP-NTFIX) or LNP-Gm-Csf mRNA-treated animals. C. Frequency and representative flow cytometry plots of CD11b+ populations in total CD45+CD3- populations. D. Expression (represented by MFI) of CD80, CD86, CD40 and class II I-A/I-E on gated total CD11c+ cells. Results shown are represented as mean +/− SEM (n = 4–5) (A-C) and two (D) independent experiments. Adjusted *P-value < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance was calculated using one-way ANOVA.

Following flow cytometric evaluation and confirmation of Treg induction and myeloid expansion, mice were treated for 4 days with Gm-csf mRNA followed by MPTP intoxication to determine its effect on neuroprotective activities. After MPTP intoxication, the total number of surviving nigral dopaminergic (TH+/Nissl+) neuronal cell bodies was assessed along with their striatal projections (Figure 4A). Numbers of surviving dopaminergic neurons were significantly decreased from 10771 ± 356 to 4893 ± 489 when treated with MPTP compared to PBS controls (Figure 4B). As expected, NTFIX treatment did not lead to neuronal sparing. However, treatment with increasing doses of Gm-csf mRNA resulted in a 21% (7162 ± 327), 34% (8576 ± 494), and 36% (8758 ± 291) increase in dopaminergic neurons compared to MPTP alone. All Gm-csf mRNA doses yielded significantly higher TH+ neuronal counts than MPTP intoxication alone. Of note, treatment with 0.1 mg/kg Gm-csf mRNA was not significantly different from non-lesioned controls, supporting the potent neuroprotective potential of Gm-csf mRNA. As expected, numbers of non-dopaminergic (TH-/Nissl+) neurons remained unchanged regardless of treatment due to their lack of MPTP susceptibility [32, 38]. However, although striatal termini TH density was increased with Gm-csf mRNA treatment over MPTP, none were significantly higher, indicating increased termini susceptibility to MPTP intoxication (Figure 4C).

Figure 4. Gm-csf mRNA attenuates MPTP-induced nigrostriatal neurodegeneration and microglial activity.

Mice were treated with PBS, MPTP, or MPTP and 0.001 mg/kg Gm-csf mRNA, 0.01 mg/kg Gm-csf mRNA, 0.1 mg/kg Gm-csf mRNA, or 0.1 mg/kg NTFIX. A. Dopaminergic (TH+/Nissl+) neurons within the substantia nigra (SN) and TH+ cell termini within the striatum (STR) of mice 7 days post MPTP (SN TH+ = scale bar, 500 μm, STR = scale bar, 1000 μm). B. Stereological quantification of total numbers of surviving TH+/Nissl+ and non-dopaminergic (TH-/Nissl+) neurons within the SN following MPTP intoxication. Differences in means (± SEM, n = 4–8) were determined where P < 0.05 compared with groups treated with ^aPBS, ^bMPTP, or ^cNTFIX. Mean percent remaining total neuron number is indicated on each treatment bar. C. Relative fold change of TH density in striatal dopaminergic termini normalized to PBS controls. Differences in means (± SEM, n = 4–8) were determined where P < 0.05 compared with ^aPBS. D. Mac-1+ microglia in the SN 2 days post MPTP (scale bar, 500 μm; inset image = 200x). E. Quantification of reactive microglia within SN. Differences in means (± SEM, n = 4–6) were determined where P < 0.05 compared with ^aPBS and ^bMPTP.

Next, changes in the MPTP-induced inflammatory response were investigated. Two days after MPTP administration, a time of peak inflammation and rate of neuronal death, brains were harvested to assess reactive microglia as determined by Mac-1 expression and amoeboid morphology (Figure 4D) [39]. MPTP increased numbers of reactive microglia, elevating counts from 0 cells/mm² for PBS control mice to 104 ± 5 cells/mm² after MPTP intoxication (Figure 4E). NTFIX treatment resulted in a similar inflammatory response to MPTP alone, yielding 126 ± 8 cells/mm², indicating its lack of protective capacity. Treatment with ascending doses of Gm-csf mRNA (0.001 mg/kg to 0.1 mg/kg) decreased reactive microglial counts to 85 ± 4, 92 ± 5, and 85 ± 10 cells/mm², respectively. However, only treatment with 0.001 and 0.1 mg/kg Gm-csf mRNA resulted in a statistically significant attenuation of microglial responses.

We next sought to determine potential mechanisms for neuroprotective and anti-inflammatory profiles afforded by Gm-csf mRNA treatment. To do so, CD4+CD25+ T cells from the spleens of Gm-csf mRNA-treated donors were isolated and adoptively transferred into MPTP-intoxicated recipient mice (Figure 5). Previously isolated CD4+CD25+ T cells demonstrated high levels of FOXP3 intracellular staining affirming their Treg phenotype and immunosuppressive activities [6, 40]. After seven days post-transfer, ventral midbrains and striatum were assessed for TH+ dopaminergic neuron survival (Figure 5A). MPTP intoxication significantly reduced neuron numbers from 7065 ± 878 for PBS controls to 4042 ± 547 (Figure 5B). Although not significant, adoptive transfer of CD4+CD25+ cells from mice treated with 0.1 mg/kg Gm-csf mRNA increased TH+/Nissl+ neuron survival to 6098 ± 1115 compared to MPTP intoxication alone. However, CD4+CD25+ cells isolated from 0.01 mg/kg-treated mice did not enhance the level of neuronal survival above the MPTP lesion. Likewise, densitometric analysis of striatal termini was not altered after adoptive transfers (Figure 5C).

Figure 5. Neuroprotective responses following adoptive transfer of Gm-csf mRNA-induced Tregs.

A. Representative images of TH+/Nissl+ dopaminergic neurons within the substantia nigra (SN, top row) and projections into the striatum (STR, bottom row) of recipient mice treated with PBS or MPTP followed by adoptive transfer of Tregs that were isolated from donor mice treated with either 0.01 mg/kg Gm-csf mRNA or 0.1 mg/kg Gm-csf mRNA (SN TH+ = scale bar, 500 μm, STR = scale bar, 1000 μm). B. Quantification of total numbers of surviving dopaminergic (TH+/Nissl+) and non-dopaminergic (TH-/Nissl+) neurons within the SN following MPTP intoxication and adoptive transfer of 1 × 10⁶ Tregs. C. Densitometry analysis of TH+ cell termini within the STR with MPTP intoxication followed by adoptive transfer of Tregs. Differences in means (± SEM, n = 6–7) were determined where P < 0.05 compared with ^aPBS.

Gm-csf mRNA treatment induces a treatment-specific immunosuppressive Treg population.

Next, to assess the ability of Gm-csf mRNA to not only enhance Treg frequency and function but alter the transcriptomic phenotype of the overall CD4+ T cell population, animals were treated with either PBS or 0.1 mg/kg Gm-csf mRNA. CD4+ T cells were isolated and analyzed for distinct transcriptomic profiles using single-cell RNA sequencing (Figure 6). A mean of 1693 genes per cell for PBS-treated mice and 1871 genes per cell for Gm-csf mRNA-treated mice were detected. A total of 53,000 reads per cell in approximately 7,500 cells from both PBS- and Gm-csf mRNA-treated samples were quantified. Samples were pooled for initial unsupervised cluster analysis to allow for the identification of potentially novel cell populations and/or phenotypes resulting from Gm-csf mRNA treatment. Clusters were then mapped and filtered to display only CD3+CD4+ cell subsets (Figure 6A). Initially, using UMI differential expression patterns and unbiased cluster mapping, PBS- and Gm-csf mRNA-treatments resulted in similar cell clustering profiles within the overall sample. Five clusters representing distinct transcriptional T cell phenotypes were displayed. Using expression levels of known cell surface markers, transcription factors, and cytokine production markers, cluster phenotypes were identified (Supplemental Figure 2). Clusters 1, 2, and 4 were identified as largely CD44-CD62L+ naïve T cell (Tn) populations. Cluster 4 also exhibited a non-significant increase in Th9 transcript expression. Cluster 3 displayed a Treg, Th2, and Th22 phenotype, and Cluster 5 exhibited markers known to be associated with Th1 and Th17 subsets. All clusters were observed in both PBS- and Gm-csf mRNA-treated samples. However, aggregation analysis identified a cluster of T cells more specific to Gm-csf mRNA treatment compared to PBS treatment alone (Figure 6B), and corresponded to Cluster 3 (Figure 6A). This cluster exhibited over 50 key genes with significantly higher expression levels that includes gene encoding for Il2ra, Foxp3, Ccr4, Il10, Ctla4, Nrp1, Nrn1, Cd83, Lrrc32, Lag3, Ikzf2, Il9r, Areg, Il1rl1, Zbtb32, Hopx, s100a4, s100a6, and s100a10 (Figure 6C), thus suggesting the induction of an anti-inflammatory phenotype by treatment with Gm-csf mRNA.

Figure 6.

Gm-csf mRNA treatment induces a highly suppressive Treg phenotype. A. t-Distributed Stochastic Neighbor Embedding (tSNE) visualization of CD3+CD4+ T cell populations based on UMI (Unique Molecular Identifier) differential expression profiles identified through unsupervised cluster analysis via single-cell RNA sequencing. Clusters are color-coded. B. tSNE aggregation analysis of PBS- (blue dots) and 0.1 mg/kg Gm-csf mRNA-treated (orange dots) samples for treatment-induced population identification. A Gm-csf mRNA treatment population is defined by the black oval. C. Key feature profile of Gm-csf mRNA-specific Cluster 3 (green in panel A) as determined by log2-fold change (L2FC) values and significant p-values. The L2FC is the ratio of the normalized mean gene UMI counts in each cluster relative to all other clusters. Key features used to map clusters were identified with UMIs >2.0 and p-values >0.001.

Gm-csf mRNA treatment leads to neuroprotective and anti-inflammatory responses in an alpha-synuclein (α-Syn) rat PD model.

To confirm our prior observations in a second species and more chronic and progressive disease model, we assessed the effect of LNP-formulated rat Gm-csf mRNA in rats. Initially, naïve rats were treated for 4 days with rat Gm-csf mRNA at a dose of 0.01 or 0.1 mg/kg. Both mRNA doses resulted in the same spleen enlargement observed in mice (Figure 7A). Flow cytometric analysis of T cell populations within total lymphocytes in peripheral blood indicated no change in CD3 percentage with any treatment (Figure 7B), a significant decrease in CD4 percentage with 0.1 mg/kg Gm-csf mRNA treatment (Figure 7C), and a significant increase in CD4+CD25+FOXP3+ Treg percentages with both mRNA doses (Figure 7D). Treg frequencies within the CD4+ population were increased from 4.1 ± 0.26 % in control rats to 12 ± 1.5% in 0.01 mg/kg Gm-csf mRNA-treated rats, and 20 ± 2.4 % in 0.1 mg/kg Gm-csf mRNA-treated rats, mimicking the observations in mice. CD4+CD25+ Tregs enriched from each treatment group were assessed for changes in their suppressive ability to inhibit proliferation of CD4+CD25- Tresps (Figure 7E-F). Treatment with both dosages of Gm-csf mRNA resulted in enhanced suppressive function compared to controls (R² ≥ 0.87, P < 0.01).

Figure 7.

Gm-csf mRNA significantly increases spleen size, CD4+CD25+FOXP3+ Treg cell frequencies and function in naïve and α-Syn overexpressed Sprague-Dawley rats. A. Quantification of spleen weight normalized to body weight following treatment with PBS, 0.01 mg/kg, or 0.1 mg/kg rat Gm-csf mRNA. Differences in means (± SEM, n = 3) were determined where P < 0.05 compared with ^a0 mg/kg. B-D. Flow cytometric analysis of T cell phenotype frequencies including CD3+ (B), CD4+ (C), and CD4+CD25+FOXP3+ (D) subsets in peripheral blood following treatment. Differences in means (± SEM, n = 3) were determined where P < 0.05 compared with ^a0 mg/kg. E. Table of mean % Tresp Proliferation ± SEM and mean % Treg inhibition for each treatment dose and Tresp:Treg ratio. F. Quantification of Treg-mediated cell suppression (± SEM) at various Tresp:Treg ratios. Treg-mediated suppression is expressed as percent inhibition. Linear regression analysis indicates P < 0.01, R² ≥ 0.87 for all treatments. G-J. Flow cytometric analysis of T cell phenotype frequencies pre- (black bars) and post- (gray bars) treatment with either Sham, AAV2/1-GFP (AAV-GFP) vector, AAV2/1-α-Syn (AAV-α-Syn) vector, AAV-α-Syn + 0.01 mg/kg Gm-csf mRNA, or AAV-α-Syn + 0.1 mg/kg Gm-csf mRNA. Peripheral whole blood was analyzed for the frequency of CD3+ (G), CD8+ (H), CD4+ (I) in within the lymphocyte population, or CD4+CD25+FOXP3+ (J) cells within the CD4+ population. Differences in means (± SEM, n = 7) were determined where P < 0.05 compared with *pre- or post-treatment with ^aSham, ^bAAV-GFP, ^cAAV- α-Syn, or ^dAAV-α-Syn + 0.01 mg/kg Gm-csf mRNA.

Due to the ability of rat Gm-csf mRNA treatment to selectively increase Treg frequency and function in healthy naive animals, we next assessed the effects of Gm-csf mRNA in a Parkinsonian model utilizing human α-Syn overexpression. Rats were administered a unilateral stereotactic injection of PBS (Sham), AAV-GFP (Vector control), or AAV-α-Syn, followed by Gm-csf mRNA administered i.m. the first 4 days and then every other day thereafter. After 28 days, T cell levels were assessed in peripheral blood (Figure 7G-J). Levels of CD3+ cells were significantly decreased from pretreatment baseline in animals receiving AAV-α-Syn + 0.1 mg/kg Gm-csf mRNA but were not significantly altered by other treatments (Figure 7G). Levels of CD8+ and CD4+ cells were also not significantly affected by any treatment; however, a mild decreasing trend was observed with mRNA treatment (Figure 7H and 7I). Recapitulating observations in mice, Gm-csf mRNA treatment increased CD4+CD25+FOXP3+ Treg frequencies within the CD4+ population of peripheral blood (Figure 7J). Treatment with 0.01 and 0.1 mg/kg Gm-csf mRNA in α-Syn overexpressing animals significantly increased Treg frequencies from 2.5 to 4.0 and 7.4%, respectively.

We also assessed the neuroprotective capacity of Gm-csf mRNA in □-syn overexpressing rats. AAV-α-Syn injection led to a remarkable diminution of dopaminergic neurons as determined by loss of TH expression compared to Sham- or AAV-GFP-treated animals (Figure 8A). Overexpression of α-Syn markedly reduced TH+ neuron counts by 49.3%. However, treatment with Gm-csf mRNA rescued the neuronal loss by sparing 57% of the neurons (28.2% and 28.8% losses) regardless of dose, thus demonstrating their neuroprotective potential (Figure 8B). Similarly, treatment with 0.01 and 0.1 mg/kg of Gm-csf mRNA also increased dopaminergic termini survival, resulting in a 13.7% and 27.9% level of protection, respectively compared to AAV-α-Syn + PBS treatment (Figure 9C). We also assessed levels of reactive microglia as Iba-1+ amoeboid microglia within the substantia nigra (Figure 8D). Administration of control AAV-vector did not elevate reactive microglial ratios over Sham injection, whereas overexpression of α-Syn increased reactive microglia by 3-fold (Figure 8E). Treatment with 0.01 and 0.1 mg/kg of Gm-csf mRNA attenuated the α-Syn-associated microglial response with fold ratios diminished to 1.96 ± 0.22 and 2.28 ± 0.09, respectively.

Figure 8.

Chronic Gm-csf mRNA treatment is neuroprotective in an α-Syn overexpres sion rat PD model. A. Representative images of TH+/Nissl+ dopaminergic neurons within the substantia nigra (column 1 and 2) of Sprague-Dawley rats that were stereotaxically-injected on the ipsilateral side with either PBS (Sham), an AAV control (AAV-GFP), AAV-α-Syn alone, or AAV-α-Syn followed by treatment with two different doses of Gm-csf mRNA, 0.01 mg/kg or 0.1 mg/kg (scale bar, 1000 μm). Representative images of TH+ dopaminergic neuron termini within the striatum after treatment are displayed in column 3 (scale bar, 1000 μm). B. Stereological quantification of the ipsilateral/contralateral ratios of total numbers of surviving dopaminergic (TH+/Nissl+, black bar) and non-dopaminergic (TH-/Nissl+, grey bar) neurons within the ipsilateral and contralateral hemispheres of the SN following α-Syn overexpression. Differences in means (± SEM, n = 7) were determined where P < 0.05 compared with ^aSham, ^bAAV-GFP, or ^cAAV-α-Syn treatment. C. Ipsilateral/contralateral ratios of striatal TH dopaminergic termini density within ipsilateral and contralateral hemispheres of the striatum. Differences in means (± SEM, n = 7) were determined where P < 0.05 compared with ^aSham, ^bAAV-GFP, ^cAAV-α-Syn, or ^dAAV-α-Syn + 0.01 mg/kg Gm-csf mRNA. D. Representative images of Iba1+ microglia within the substantia nigra on both contralateral and ipsilateral sides (scale bar, 40 μm). E. Quantification of reactive, amoeboid Iba1+ microglia density utilizing stereological analysis displayed as a ratio of ipsilateral and contralateral densities. Differences in means (± SEM, n = 7) were determined where P < 0.05 compared with ^aSham, ^bAAV-GFP, or ^cAAV-α-Syn. F-I. Quantification of pro- and anti-inflammatory cytokine profiles within plasma. Fold change with treatment is relative to AAV-α-Syn alone. F. Fold change of CINC-1, CINC-2 α/β, CINC-3, CNTF, Fractalkine, GMCSF, siCAM-1, and IP-10. G. Fold change of IFNy, IL-1α, IL-1Β, IL-1ra, IL-2, IL-3, and IL-4. H. Fold change of LIX, L-Selectin, MIG, MIP-1α, RANTES, CXCL7, and TIMP-1. I. Fold change of IL-6, IL-10, IL-13, IL-17, TNFα, and VEGFA. Differences in means (± SEM, n = 4) were determined where P < 0.05 compared with ^aAAV-α-Syn.

Along with microglial responses, cytokine protein levels within plasma were also assessed following 28 days of AAV-α-Syn overexpression to determine changes in the inflammatory response associated with disease (Figure 8F-I). Treatment with 0.01 mg/kg Gm-csf mRNA resulted in decreases in CINC-1, CINC-2αβ, CINC-3, CNTF, IP-10, LIX, IL-1ra, IL-2, and TNFα (p = 0.09). Treatment with 0.1 mg/kg Gm-csf mRNA resulted in decreased CINC-1, CINC-2αβ, CINC-3, CNTF, IP-10, LIX, MIP-1α, RANTES, IL-1α, IL-1ra, IL-2, and TNFα relative to levels observed in AAV-α-Syn overexpression alone. Proteins found to be increased in both treatment doses were CXCL7, TIMP-1, IFNγ, and IL-10. However, given these observations, only levels of LIX were significantly downregulated from AAV-α-Syn overexpression alone.

Discussion and Conclusions

GM-CSF protein treatment is neuroprotective in neurodegenerative disorders [3, 41–44]. This occurs, in part, by induction of Tregs leading to profound control of innate microglial immune responses. The balance between phenotype-specific Teff-mediated neurodegeneration or Treg-mediated neuroprotection is posited to result in immunomodulation and disease control [13, 45–49]. In support of this idea, adoptive transfer of CD4+CD25+ Treg into MPTP-intoxicated animals attenuates neuroinflammation leading to nigrostriatal protection [4]. We, and others, have shown that GM-CSF protein readily increases Treg numbers and activity with parallel neuroprotective activities in PD, AD, myasthenia gravis, and traumatic brain injury (TBI) models of human disease [3, 6, 7, 10, 41, 42, 44]. However, allergic treatment site reactions, muscle and bone pain, and humoral responses targeted against the protein are all side effects associated with its daily clinical use [10, 25, 26, 29]. In attempts to overcome untoward reactions and increase potency of the agent, a Gm-csf mRNA delivery strategy was developed to produce GM-CSF protein in situ, leading to parallel white blood cell increases with a selective Treg induction. No major untoward metabolic effects were observed, suggesting that Gm-csf mRNA is potentially safe for clinical use as demonstrated by other mRNA delivery methods [50]. However, increased spleen size was noted in both mice and rats following treatment, and is a well-known phenomenon following GM-CSF treatment [51]. GM-CSF is used to reconstitute myeloid-derived immune populations, which can occur by extramedullary hematopoiesis within the spleen in young rodents [52]. Therefore, spleen enlargement was likely to be observed following treatment due to the rapid immune reconstitution. Likewise, Gm-csf mRNA treatment led to classical expansion of myeloid populations, specifically dendritic cells (DCs). DCs can adopt multiple maturation states with control over many pathways [53]. Therefore, we explored several canonical markers of DC activation. Indeed, treatment led to increased CD80 and CD86 expression within the DC population. GM-CSF has been shown to upregulate CD80 and CD86 under certain circumstances [54, 55]. Thus, we believe that treatment duration and level of exposure are responsible for the upregulation observed. Overall, our observations are consistent with an intermediate phenotype, suggesting a semi-mature DC population. Partial DC maturation has been linked to immune tolerance and Treg production [56], whereas full maturation is linked to immunogenicity and signaling to advance effector T cell differentiation [53]. We also observed altered alkaline phosphatase, albumin, and amylase levels with the highest treatment group. However, these alterations still fall within normal rodent ranges and are not concerning [57–59].

LNP containing Gm-csf mRNA was administered prior to acute MPTP intoxication or during chronic exposure to α-Syn overexpression and resulted in neuroprotective responses with enhanced neuronal survival and attenuation of microglial-associated inflammation. Likewise, the neuroprotective phenotype from adoptive transfer of Gm-csf mRNA-induced Tregs supports the abilities of this cell type to directly attenuate the neuroinflammatory cascade in brain. Gm-csf mRNA treatment also shifted peripheral CD4+ T cells into an immunosuppressive phenotype and function contributing to its neuroprotective activities. These results are similar to those previously achieved through direct administration of recombinant GM-CSF protein or adoptive transfer of GM-CSF-induced Tregs [6]. The resulting phenotype shift has also been noted while treating PD subjects with sargramostim (GM-CSF, Leukine) (NCT03790670) [10]. Notably, we were able to successfully achieve similar neuroprotective effects using this alternative delivery approach providing the potential to decrease the frequency of adverse events associated with recombinant protein administration. Apart from determining the potential safety of its use and ability to modulate the immune response, treatment with lower doses of Gm-csf mRNA also provided the ability to forgo daily treatment. In our rat studies, prolonging Gm-csf mRNA treatment to every other day resulted in significant neuroprotective and anti-inflammatory responses. This is likely due to the highly elevated plasma concentration of GM-CSF protein following treatment. Therefore, this formulation acts as a proof-of-concept for the development of longer-acting mRNA formulations that will extend dosing intervals even further with the hope of decreasing known side effects associated with daily injection.

The neuroprotective capacity of Gm-csf mRNA is certainly linked to modulation of both innate and adaptive immunity which affect both pro- and anti-inflammatory mediators in PD models. Not only did treatment selectively increase Treg numbers and function in multiple species, but also resulted in increased anti-inflammatory responses and induced a CD4+ T cell population with an anti-inflammatory phenotype. Many of the key features identified using cluster analysis were linked to basic biological cell functions such as cell adhesion, trafficking, migration, and proliferation. However, nearly half of the observed features can be directly linked to a highly immunosuppressive and anti-oxidative Treg phenotype, including Il2ra, Foxp3, Ccr4, Il10, Ctla4, Nrp1, Nrn1, Cd83, Lrrc32, Lag3, Ikzf2, Il9r, Areg, Il1rl1, Zbtb32, Hopx, s100a4, s100a6, and s100a10 (Figure 6C). Classically, Il2ra, Foxp3, Ccr4, Il10, and Ctla4 encode for known Treg phenotype markers responsible for immunosuppressive mechanisms including IL-2 sequestration, IL-10 cytokine production, and inhibition of antigen presentation [60]. The significant upregulation in these markers confirms the presence of an induced Treg phenotype. Elevated levels of additional surface marker genes including Nrp1, Nrn1, Cd83, Lrrc32, and Lag3 correspond with Treg differentiation, survival, and stability [61–65]. Additionally, significant upregulation in gene markers associated with enhanced immunosuppressive function were also observed, including Ikzf2, Il9r, Areg, and Il1rl. Ikzf2 encodes for the zinc finger protein Helios, and its presence is linked to a more immunosuppressive phenotype compared to Helios- Treg [65]. Likewise, Tregs expressing the IL9 receptor, amphiregulin, and ST2 also have enhanced immunosuppressive capacity via increased IL-9 signaling and the ST2/IL-33 signaling axis [66–68]. ST2+ Treg are highly activated and maintain their suppression through increased IL-10 production and TGFβ release [68]. Gm-csf mRNA treatment also generated an induced Treg (iTreg) phenotype through the upregulation of Zbtb32 and Hopx [69, 70]. Both genes are linked to tolerance induction and innate and adaptive immune cell interactions. Lastly, proteins encoded by s100a4, s100a6, and s100a10 are present in the defined Treg phenotype and have potent antioxidant functions for immune cell regulation [70, 71][72]. Taken together, these key features can be linked to the generation of an enhanced immunosuppressive Treg population following Gm-csf mRNA treatment. Further subset analysis and more distinct clustering based on differential expression levels and K means will be carried out in future studies to identify more extensive cell phenotypes within the defined clusters linked to Gm-csf mRNA treatment.

Similarly, in the rat PD model, Gm-csf mRNA treatment led to decreases in CINC proteins needed for neutrophil attraction and proteins associated with macrophage activation, recruitment, and initiation of a pro-inflammatory immune response such as LIX, IP-10, MIP-1α, IL-1α, IL-2, and RANTES [73–75]. On the other hand, Gm-csf mRNA treatment increased IL-10 levels, a potent anti-inflammatory response mediator that decreases the immune response [60]. Taken together, observed gene and protein changes within the periphery likely are responsible for the decreased inflammatory microglial response observed within the brain in both the MPTP mouse model and α-Syn overexpression model. Previously, peripheral immune effects were shown to have profound neuroprotective effects, even without penetration of modulatory agents across the BBB [4–7]. Both T cells and cytokines can cross the BBB, potentially leading to indirect neuroprotective effects. Specifically, we have shown that Tregs migrate into the brain in limited numbers [76]. Reports by others have affirmed these results by demonstrating Treg motility for laminin migration. These lymphocyte populations show highly mobility potential than other T cell subsets in trans-well assays as seen through microporous membrane tests [77]. Results also demonstrated migration across the brain endothelium, wherein human Tregs from healthy donors show rapid migratory abilities across primary human brain endothelium. Therefore, GM-CSF-induced Tregs may have the capacity to penetrate the brain and affect cell-cell interactions within the brain microenvironment. Alternatively, direct GM-CSF neuroprotective activities remains possible, given the cytokine’s ability to cross the BBB. These studies should be explored to fully understand the underlying neuroprotective mechanisms achieved. Relevant to this study conclusions rests in the fact that axonopathy is an early neurodegenerative sign [78] and that MPTP strongly affects the striatum [5–7, 40]. GM-CSF is known to protect the dopaminergic striatal termini and is underscored by increases in TH+ densities by GM-CSF. Also, GM-CSF slows neuronal loss while protecting the striatum, as seen in α-Syn overexpressing rats. Thus, the GM-CSF effect is neuroprotective both for dopaminergic cell bodies and their projections into the striatum.

Overall, we posited that treating with Gm-csf mRNA would lead to robust neuroprotective outcomes similar to treatment with recombinant GM-CSF protein. While Gm-csf mRNA increases Treg numbers and function, it also reduces reactive microgliosis, enhances neuronal survival, and induces a profound phenotypic shift of the CD4+ T cell phenotypic repertoire. These findings were achieved using an alternative delivery approach, which decreased the dosage concentrations and prolonged the dosing intervals affording the potential for decreased side effects currently associated with treatment in existing clinical dosing schemes [10, 25, 26, 28, 29]. Also, it is now widely accepted that GM-CSF treatment induces neuroprotective Treg for stroke, PD, ALS and AD [1–3, 13, 44, 47, 48, 79–81]. Likewise, its neuroprotective nature across species and in both an acute MPTP model and chronic α-Syn overexpression model strengthens its clinical relevance and potential clinical utility. Treatment of PD subjects with recombinant GM-CSF has already indicated the cytokine’s clinical efficacy, safety, tolerability, and impact on improving motor function (NCT03790670) [10]. It is likely that the types of alternative deliveries suggested in this report would further improve clinical outcomes. Therefore, taken together, this study serves to support the utilization of Gm-csf mRNA delivery as a potential therapeutic approach for PD and other neurodegenerative diseases. Lastly, given the increasing interest in Treg-mediated therapies for neurodegenerative and autoimmune disease, a potential increase in cancer risk remains a concern. Tregs can accelerate tumor progression, but do not initiate disease [82]. For PD patients, decreased Treg numbers and immunosuppressive function were recorded [8]. On balance, Treg-enhancing or -inducing therapies used to restore levels similar to their healthy, age-matched counterparts would be beneficial to slow disease and would pose limited ties to cancer development.

Supplementary Material

1. Supplemental Figure 1. Experimental design for MPTP intoxication and alpha synuclein (α-Syn) overexpression studies.

The top panel depicts the Gm-csf mRNA pretreatment followed by MPTP intoxication study design and timeline. The middle panel depicts the Treg adoptive transfer timeline, and the bottom panel depicts the experimental scheme for the AAV-α-Syn overexpression studies in rats. Experimental treatment groups, n values, age, and sex of animals are indicated for each study design and are located in their corresponding panels.

2. Supplemental Figure 2. Identification of T cell phenotype using unbiased cluster analysis and known T cell markers.

Tables depicted here represent common T cell marker genes that correspond with cell surface markers, transcription factors, and cytokine profiles for Th1, Th2, Th17, Th22, Th9, Treg, and/or Tn. Tables are color-coded to match the defined clusters in Figure 6A. Differential expression levels are reported as Log2 Fold Change (L2FC) along with the corresponding p values. The L2FC is the ratio of the normalized mean gene UMI counts in each cluster relative to all other clusters.