Farnesol induces protection against murine CNS inflammatory demyelination and modifies gut microbiome

Abstract

Farnesol is a 15-carbon organic isoprenol synthesized by plants and mammals with antioxidant, anti-inflammatory, and neuroprotective activities. We sought to determine whether farnesol treatment would result in protection against murine experimental autoimmune encephalomyelitis (EAE), a well-established model of multiple sclerosis (MS). We compared disease progression and severity in C57BL/6 mice treated orally with 100 mg/kg/day farnesol solubilized in corn oil to corn-oil treated and untreated EAE mice. Farnesol significantly delayed the onset of EAE (by ~2 days) and dramatically decreased disease severity (~80%) compared to controls. Disease protection by farnesol was associated with a significant reduction in spinal cord infiltration by monocytes-macrophages, dendritic cells, CD4⁺ T cells, and a significant gut change microbiota composition, including a decrease in the Firmicutes:Bacteroidetes ratio. The study suggests FOL could protect MS patients against CNS inflammatory demyelination by partially modulating the gut microbiome’s composition.

1. Introduction

Over the last decade, a reciprocal association between gut microbiota and central nervous system (CNS) diseases has been proposed. Among CNS diseases, disorders characterized by inflammatory demyelination such as multiple sclerosis (MS) are under extensive scrutiny. Multiple sclerosis is an autoimmune disease that affects approximately 2 to 3 million, with a 50 – 300 per 100,000 prevalence [1]. It is a complex disorder with uncertain pathogenesis. However, it is well established that in MS, immune cells attack axonal myelin, causing demyelination, axonal loss, and neuronal death [2]. The disease presents with a wide array of symptoms, including muscle weakness, numbness, loss of coordination, blurring or double vision, pain, or paralysis, among others [3]. Pro-inflammatory T helper 17 (Th17) cells and Th1 cells have been proposed as central pathogenic populations driving the autoimmune reactions in MS and MS’s animal models. By contrast, regulatory cell populations such as regulatory T cells (Tregs) and the anti-inflammatory cytokines they secrete, such as interleukin-10 (IL-10), may play a role in controlling disease [2].

Most studies have shown that relapsing MS patients’ intestinal microbiome is not significantly different from healthy individuals’ microbiome when comparing alpha and beta diversities. However, significant changes in specific bacterial and archaeal genera and species have been observed [4–8]. Further, experimental evidence gathered in the most widely used murine models of MS, the experimental autoimmune encephalomyelitis (EAE) and the Theiler’s murine encephalomyelitis virus (TMEV) models, has shown that complete microbial absence in gnotobiotic animals [9,10], alterations of the microbiome with antibiotics [11–13], and treatments with specific bacterial strains [14–16] or microbial products [17–19], all impact the progression of the disease. Furthermore, the fecal transplantation from MS individuals to gnotobiotic mice with reduced EAE severity restores disease severity by negatively impacting the function of IL-10-producing regulatory T cell populations [4,5].

Farnesol (FOL), a 15-carbon organic acyclic sesquiterpene primary alcohol produced by plants and animals from 5-carbon isoprene precursors and dephosphorylation of farnesyl pyrophosphate is a key intermediate of the cholesterol synthesis pathway. Both cis,trans (plants), and all-trans (mammals) FOL isomers are potent pharmacological antagonists of L- and N-type voltage-gated Ca²⁺ channels, channels implicated in cell growth and neurotransmission [20–23]. Whether endogenous FOL regulates the activity of these channels in vivo remains to be established. However, the presence of FOL in the human brain suggests that FOL is a signaling molecule potentially involved in regulating CNS calcium homeostasis, neurotransmission, and protection against neuronal calcium overload [24]. FOL also has potent anti-oxidant and anti-inflammatory effects in vitro [25,26]. Neuroprotection by FOL was demonstrated in a murine model of neurodegeneration (LPS-induced) and a model of neurotoxicity (acrylamide-induced) in a mechanism based on the regulation of the production of free radicals by glial cells and pro-inflammatory cytokines production in the CNS [27,28]. In these studies, FOL was administered intraperitoneally (i.p.) at a daily dose of 100 mg/kg without evidence of toxicity while improving gait performance, neuromuscular function, and fine motor coordination. Because of the proposed anti-inflammatory and neuroprotective features described for FOL, we hypothesized that oral treatment with the isoprenol would protect against EAE and improve autoimmunity markers. We tested our hypothesis in mice treated daily with FOL by gavage from EAE-induction day to end of study (day 26). The treatment’s impact on clinical severity was recorded daily, whereas the effects on immunity were assessed on day 26 by flow cytometry of spinal cord lymphocytes.

Farnesol has also been shown to inhibit the formation of biofilm by fungal strains of Candida albicans [29–31] and the Gram-positive human pathogenic strain methicillin-resistant Staphylococcus aureus (MRSA) [32,33]. The effect of the isoprenol on biofilms is associated with apoptosis induction in C. albicans [34], and disruption of cell membrane integrity in MRSA [33], while studies in Bacillus subtilis reported that FOL promotes biofilm formation [35]. Interestingly, the treatment of MS patients with immunomodulatory drugs (e.g., beta-interferon, dimethyl fumarate, or glatiramer acetate) triggers changes in intestinal microbial species [8][36]. Because the gut microbiota, in turn, modulates the immune response of the host [37], and EAE incidence and progression [4,5,9–12], we sought to determine if the treatment with FOL would favorably impact the composition of the gut microbiota. We thus collected stool samples before EAE induction (day 0), soon after disease onset (day 14), and at the end of treatment (day 26) and analyzed their microbial composition.

2. Materials and methods

2.1. Animals

Ten week-old female C57BL6 NHsd mice weighing approximately 20 g were obtained from Envigo (Envigo RMS, Inc., Indianapolis, IN, USA). The animals were housed in wire-top cages (5 animals per cage; 22 ± 1°C; 23-33 % humidity; 12-hour light/dark cycle) and had free access to food (Teklad 2018) and water. All animal care and procedures were under Eastern Washington University institutional policies for animal health and well-being.

2.2. EAE induction

Upon arrival, animals were given one week to acclimate to their new housing environment. EAE was induced using a kit from Hooke Laboratories (Hooke Kit™ EK-2110, Hooke Laboratories, Lawrence, MA). The kit includes MOG_35-55 (self-antigen) in emulsion with complete Freund’s adjuvant (CFA) and Pertussis toxin (PTX) in glycerol buffer. EAE induction was initiated (day 0) with the injection of the MOG_35-55-CFA emulsion, and the PTX diluted in phosphate-buffered saline. The toxin injection was repeated the following day (day +1).

2.3. Treatments

Two separate EAE protection studies were performed, each with 3 treatment groups: mice receiving 100 mg/kg/day FOL (trans,trans-farnesol, Sigma-Aldrich, cat.# 277541) solubilized in corn oil (FOL-EAE; n = 18), mice receiving corn oil only (CO-EAE; n = 23), and untreated (naïve) mice (U-EAE mice; n = 24). Treatments (corn oil, farnesol) were administered by gavage and adjusted weekly for the animals’ body weights. The results of these two studies have been combined (see Results, Fig. 1). A third EAE protection study was performed for flow cytometry analysis of the spinal cord (nine untreated, nine corn-oil treated, and ten FOL-treated mice per group).

Figure 1.

Oral treatment with FOL improves disease severity in EAE mice (FOL-EAE, n =18; CO-EAE, n = 23; U-EAE, n = 24). A) Clinical severity score from induction (day 0) to end of treatment (day 26) (treatments from day 0 to 26). B. Percent survival. C. Bodyweight. D. EAE onset. E. Clinical severity score at day 26. Data are presented as mean ±1SEM. Statistical analysis was performed as described in Methods. *, p < 0.05 (for U-EAE vs. FOL-EAE in panel C); ***, p < 0.001 (in panel A); #, p < 0.05 (for U-EAE vs. CO-EAE in panel C).

2.4. Weights

Body weights were measured weekly (days 0, 7, 14, 21) and at day 26 (euthanasia) and are expressed as % of body weight at the time of EAE induction.

2.5. EAE clinical scores evaluation

Mice were scored as described by us and others: 0 – no detectable signs of EAE, 0.5 – distal limp tail, 1.0 – complete limp tail, 1.5 – limp tail and hind limb weakness, 2.0 – unilateral partial hind limb paralysis, 2.5 – bilateral partial hind limb paralysis, 3.0 – complete bilateral hind limb paralysis, 3.5 - complete bilateral hind limb paralysis and partial front limb paralysis, 4.0 - quadriplegia. To ease access to water, mice were given water bottles upon the day of disease induction. To facilitate access to food, when mice exhibited a score of 2.5, food was soaked with water and left in a shallow cup placed close to where mouse bedding was located. Mice were scored daily until day 26.

2.6. Spinal cord cell preparation and flow cytometry

In order to determine the impact of the treatments on CNS infiltration by nucleated cells, spinal cord and splenic cells were isolated aseptically from mice on day 19 after disease induction, a day when some of the untreated EAE animals have reached maximum disease severity and were euthanized. Spinal cord cells were also isolated from surviving mice at the end of treatment, on day 26 after disease induction. Spinal cords were flushed out of the vertebral cavity with 1 mL cold sterile phosphate buffer saline (PBS) using a 3-mL syringe and 18-gauge needles. Extracted spinal cords were homogenized and passed through a 70-μm filter in 1 % bovine serum albumin (BSA) in sterile PBS. After centrifugation (350 x g; 5 mins) and one PBS wash, the cell pellets were suspended in sterile PBS, passed through a 40-μm filter, centrifuged (350 x g; 5 mins), and resuspended in PBS-1 % BSA. Splenocytes were obtained after homogenization of spleens with red blood cell lysis buffer (Roche Diagnostics GmbH, Mannheim, Germany, cat. # 11 814 389 001). Spinal cord cells and splenocytes were then treated with an anti-mouse CD16/CD32 Fc block and stained with hamster anti-CD11b FITC (clone M1/70; BD Biosciences, San Jose, CA, cat. #. 553310), hamster anti-CD11c (clone HL3; BD Biosciences, cat. # 553802), and rat anti-mouse F4/80 APC (clone T45-2342; BD Biosciences, cat. # 566787) for differentiating monocyte/macrophages and CD11c-positive dendritic cells. Tregs were identified using a mouse regulatory T cell staining kit from eBiosciences (cat. # 88-8111-40; Thermo Fisher Scientific, Waltham, MA) containing the following antibodies: anti-CD4 FITC (RM4-5), anti-CD25 APC (PC61.5), and anti -mouse/rat Foxp3 PE (FJK-16s). Rat anti-mouse IgG2a PE was used as isotype control for intracellular staining. Samples were analyzed using a BD Accuri C6 flow cytometer (BD Biosciences). Data processing was performed with the FloJo software (FloJo LLC, Ashland, OR).

2.7. Isolation of fecal samples and 16S ribosomal RNA (rRNA) sequencing

Stool samples were collected in sterile tubes on days 0, 14, and 26, stored at −80°C and sent to AKESOgen (Norcross, GA) for 16S rRNA analysis of the microbiome. Qiagen DNA stool extraction kits were used for DNA isolation. DNA aliquots (1 ng/ml DNA) were analyzed by PCR using primers specific to the variable region 4 (V4) of the prokaryotic 16S rRNA gene. Library preparation and sequencing for V4 amplicon sequencing were performed on the Illumina MiSeq platform. A modified protocol with Nextera XT kit was used for library preparation, and sequencing was performed using MiSeq V2 (2x250bp) chemistry. AKESOgen used a protocol that combined the 2-steps in 1-step of amplification with forward and indexed-reverse primers. After sequencing, the microbiome data were analyzed using Nephele, a cloud-based web application from the Office of Cyber Infrastructure and Computational Biology (OCICB), National Institute of Allergy and Infectious Diseases (http://nephele.niaid.nih.gov/; 2016). QIIME was used for analysis [38] and R for statistical analysis [39]. The Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) package was also used with Nephele. A total of 83 samples were compared using the QIIME FASTQ paired-end protocol. Reads that were demultiplexed were clustered into OTUs open reference approach by comparison with the SILVA_99 database allowing sequences clustered at 99% similarity. Analyses included the identification of chimeras and removal using uchime. The abundance of each taxon was analyzed using the phyloseq package in R [39]. The compositional heterogeneity of each sample’s microbial community at every time point and each taxonomic level was visualized using PCoA scaling and the ordinate function in the phyloseq package and using the Bray-Curtis dissimilarity index calculated using the metaMDS function in the vegan package.

2.8. Statistical analysis

For EAE clinical scores, body weights, and body weight changes, group differences were estimated using repeated measures and mixed-effect ANOVA followed by Tukey’s multiple comparison post-hoc test. Group differences in disease onset, severity scores, spinal cord, and spleen cell infiltration were evaluated using non-parametric Kruskal-Wallis followed by Dunn’s multiple comparisons tests. For the statistical analysis of the microbiota, the R software package was used. ADONIS was used to determine group differences in gut microbiota composition estimated by PCoA analysis (adjusted p values provided), and differences in microbiota alpha diversity and Firmicutes:Bacteroidetes (F/B) ratios were evaluated using non-parametric Kruskal-Wallis followed by Dunn’s multiple comparisons testing. The statistical analysis of the microbiota by ADONIS was complemented with a two-way ANOVA followed by Tukey’s test to assess group differences at the phylum, family, and genus level, obtaining adjusted p values when comparing all mice and timepoints combined for each taxonomical level. The statistical analysis was performed with Prism (version 9) from GraphPad Software, Inc.

3. Results

3.1. Farnesol confers protection against experimental autoimmune encephalomyelitis (EAE)

Because of the reported anti-inflammatory and neuroprotective activities of FOL on microglia and neurons, respectively, we sought to determine whether FOL oral administration would result in protection against EAE, the most widely used animal model to study MS. We chose to use the MOG_35-55-induced C57BL/6 EAE due to the model’s inflammatory and neurodegenerative features [40]. Only female mice were used in our studies, and disease was induced in 10-week-old mice a week after they arrived at our facilities.

A significant overall Treatment effect on clinical scores was observed when comparing untreated, vehicle-treated, and FOL-treated mice (Fig. 1A; p < 0.001). There was also a significant Time effect on clinical scores (p < 0.001), and a significant Treatment x Time interaction (p < 0.001). The Treatment vs. Time statistical analysis showed that EAE severity was on average significantly lower in FOL-treated mice than in untreated and corn-oil treated mice (~80%; p < 0.001; Fig. 1A). Corn oil treatment also decreased disease severity (p < 0.001), but the corn-oil effect amounted to only a ~25% reduction in severity, significantly less than that of farnesol (Fig. 1A). Disease severity was estimated in surviving animals, attributing a default severity score of 5 to animals that had to be euthanized during the treatment period. To get a more accurate estimate of disease protection by the treatments, we thus examined group differences in the percentage of animals that had to be euthanized because of their high clinical scores. This analysis showed that in the FOL-EAE group, no animals had to be euthanized. In contrast, 22% of the mice in the CO-EAE group and 50% of the untreated animals had to be terminated (Fig. 1B). The vehicle- and FOL-treated EAE mice maintained the body weights throughout the disease, while a significant reduction was observed in the untreated group (day 14) (Fig. 1C).

On average, disease onset in untreated mice was 12.6 ± 0.37 days (mean ± SEM) after EAE induction (Fig. 1D). In contrast, it was significantly delayed in the FOL-EAE group (14.9 ± 0.54; p = 0.0004). Disease onset in corn-oil treated mice was also delayed compared to U-EAE animals (14.1 ± 0.24, p = 0.003) and shorter on average than in FOL-treated mice. Still, the difference between CO-EAE and FOL-EAE animals did not reach statistical significance. Disease protection by FOL and corn oil was also observed when comparing EAE severity scores measured at day 26: FOL-EAE (0.79 ± 0.17) vs. U-EAE (2.81 ± 0.27), p < 0.0001; CO-EAE (2.01 ± 0.23) vs. U-EAE, p = 0.029; and FOL-EAE vs. CO-EAE, p = 0.0025; Fig. 1E).

Overall, the data in Fig. 1 show that both corn oil and farnesol reduce disease severity. The combination of corn oil with farnesol confers greater disease protection than corn oil alone.

3.2. Farnesol treatment reduces the percentages of immune cells in spinal cords of EAE mice

We next sought to determine whether the protective effects of FOL correlated with reduced levels of CD4⁺ T cell infiltration into the CNS of EAE mice. We first compared the CD4⁺ T cell and CD25⁺Foxp3⁺ subpopulation of CD4⁺ T cells in the animals’ spinal cord that survived one of the two experiments shown in figure 1 (untreated: n = 4; vehicle-treated: n = 9; FOL-treated: n = 8). At the end of the experiment (day 26), only the spinal cords of FOL-treated mice showed a significantly reduced percentage of CD4⁺ T cells when compared with untreated EAE mice (Supplemental Fig. 1; p < 0.05). There were no significant group differences in the percentages of CD4⁺ T cells with CD25⁺Foxp3⁺ phenotype (not shown). Similarly, there were no group differences in the percentage of CD8⁺ T cells (not shown).

Because of the smaller number of surviving animals in the U-EAE group than in the treated groups, we sought to determine spinal cord immune cell infiltration at treatment day 19 when all untreated animals were still alive. As shown in Fig. 2A, disease onset delay and disease protection by the corn oil and farnesol matched what we observed in the studies shown in Fig. 1. For the flow cytometry analyses, we first discriminated doublets and selected singlets for further analysis in all spinal cord preparations isolated from all EAE mice (untreated: n = 9; vehicle-treated: n = 9; FOL-treated: n = 10). We used a gating strategy previously published [41] (Fig. 2B). After singlet selection, we compared the CD11b, CD11c, and F4/80-positive and negative subpopulations and CD4⁺ T cells and Tregs in the spinal cords of all mice (Fig. 2B). The treatment with FOL resulted in significant reductions in the percentages of CD11b⁺F4/80⁻ and CD11b⁺F4/80^int monocyte subsets, F4/80^neg granulocyte/monocytes and F4/80^int granulocyte/monocytes, respectively, and CD11b⁺F4/80⁺ (monocyte-derived macrophages; MDM) subpopulations when compared with untreated and vehicle-treated mice (Fig. 2C). FOL-treated mice’s spinal cords also had a reduced percentage of CD11b⁺F4/80⁻CD11c⁺ monocyte-derived dendritic cells (moDCs) vs. untreated and vehicle-treated mice (Fig 2D). Further studies will be needed to define the phenotype of the cell populations affected by FOL treatment, including the granulocyte subsets identified and whether the cells are residents or infiltrate from the periphery.

Figure 2.

Oral treatment with FOL reduces monocyte/macrophage and dendritic cell infiltration of the spinal cord in EAE. CD11b⁺F4/80⁻, CD11b⁺F4/80^int, and CD11b⁺F4/80⁺ subpopulations (percentages) were measured by flow cytometry at day 19 post-EAE induction in U-EAE (n = 9), CO-EAE (n = 9), and FOL-EAE (n = 10) mice. Data are presented as mean ± 1 SEM. A. EAE scores from EAE induction (day 0) until peak of disease (day 19). B. Gating strategy used for the identification of CD11b⁺F4/80⁻ (F4/80^neg granulocyte/monocytes), CD11b⁺F4/80^int (F4/80^int granulocyte/monocytes), CD11b⁺F4/80⁺ (monocyte-derived macrophages; MDM), and CD11b⁺F4/80⁻CD11c⁺ (monocyte-derived dendritic cells (moDCs)) subpopulations (Figs. 2B and 2C), and the identification of CD4⁺ T cells (Fig. 3) [41]. C. Percentage of CD11b⁺F4/80⁻, CD11b⁺F4/80^int, and CD11b⁺F4/80⁺ subpopulations in singlets and total of the spinal cords (***, p < 0.01; **, p < 0.01). D. Percentage of CD11c-positive cells in gated CD11b⁺F4/80− cells (**, p < 0.01). The percentages shown in the plots indicate the frequencies of positive cells in pre-gated populations. See Methods for statistical analysis.

The analysis of CD4⁺ staining showed that the treatment with FOL reduced the percentages of CD4⁺ T cells in spinal cords significantly compared to the untreated and vehicle-treated mice (Fig. 3A). Remarkably, although the number of CD4⁺ T cells isolated was significantly reduced in FOL-treated mice’s spinal cords, the proportion of CD25⁺Foxp3⁺ in CD4⁺ T cells was increased (not significant change) compared with untreated mice, and significantly higher than vehicle-treated mice (p < 0.05) (Fig. 3B). No differences in the percentages of any of the cell subpopulations analyzed were observed when comparing untreated and vehicle-treated mice at the peak of EAE.

Figure 3.

Oral treatment with FOL reduces CD4+ T cells and increases Tregs infiltration in spinal cords in EAE. Percentages of CD4⁺ T cells and CD25⁺Foxp3⁺ in CD4⁺ T cells were measured by flow cytometry at day 19 post-EAE induction in U-EAE (n = 9), CO-EAE (n = 9), and FOL-EAE (n = 10) mice. Data are presented as mean ± 1 SEM. A. Percentage of CD4⁺ T cells in singlets and total of the spinal cords. B. Percentage of CD25⁺Foxp3⁺ events in gated CD4+ T cells. *, p < 0.05; **, p < 0.01). See Methods for statistical analysis.

To determine whether the treatments of EAE would affect Treg frequencies in the periphery, we performed the same analysis in single-cell suspensions prepared from the spleens of all mice used in the study. We did not observe significant group differences in splenic CD4⁺ T cell frequencies or proportion of CD25⁺Foxp3⁺ in splenic CD4⁺ T cells (Supplemental Fig. 2). A similar analysis was performed for splenic monocyte/macrophages and dendritic cells. Although there were no group differences in CD11b⁺F4/80⁻CD11c⁺ dendritic cell frequencies, there was a significant increase in the frequencies of F4/80^int granulocyte/monocytes and CD11b⁺F4/80⁺ MDMs in FOL-treated mice when compared with untreated animals. In contrast, the frequency of a F4/80^neg granulocyte/monocytes isolated from the spleens of corn-oil-treated mice was significantly reduced vs. untreated mice, while no significant differences were observed between untreated and FOL-treated mice and between the CO-EAE and the FOL-EAE groups (Supplemental Fig. 2).

In summary, the results of these flow cytometry studies show that disease protection conferred by farnesol at the peak of the disease correlates with a reduction in the presence in or infiltration of nucleated cells into the spinal cords of EAE mice. The data further suggest that the small fraction of infiltrated T cells correspond to cells with a regulatory phenotype. The effects observed in T cells and Tregs are only observed in the spinal cords and not in the periphery.

3.3. EAE treatment impacts the composition of the gut microbiome of EAE mice

The analysis of the intestinal (stool) microbiota of U-EAE, FOL-EAE, and CO-EAE mice (samples collected at days 0, 14, and 26) revealed marked [group] and [group x time] interaction differences (Fig. 4).

Figure 4.

Oral treatment with FOL modifies the diversity and composition of the gut microbiota. A. Schematic timeline of the EAE experiment and experimental detail of the sequencing analysis of the fecal microbiome of EAE mice. Stool samples were harvested from mice on days 0, 14, and 26 post-EAE induction. On day 26, only EAE survivors were compared. Samples were analyzed by amplifying the V4 region of the 16S rRNA and reads compared to determine alpha diversity, and beta diversity, microbiome composition at every taxonomic level. B. Alpha diversity quantified by Shannon index and Chaol index. Data are shown as mean ± minimum and maximum and compared by Kruskal-Wallis analysis followed by Dunn’s multiple comparisons test (*, p < 0.05; ***, p < 0.001). Impact of EAE induction and treatment on the composition of the intestinal microbiota. C. PCoA analysis for the genus taxonomic level of OTUs identified in samples obtained from untreated EAE mice (EAE), vehicle-treated EAE mice (VEH), and FOL-treated EAE mice (FOL) at days 14 and 26. Statistical analysis was performed using ADONIS (adjusted p values shown in tables; see Methods).

We first evaluated whether disease induction and treatment with FOL or vehicle alone would impact the intestinal microbiotas’ alpha diversity during the experiment (Fig. 4). We performed this analysis by comparing the mean ± maximum and minimum of Shannon index, and Chao1 index (Fig. 4B) detected in stool samples of animals grouped by treatment for days 0, 14, and 26. Each treatment group consisted of a maximum of three cages with 4 animals per cage or less by the end of the experiment. There was no significant group difference in the Shannon and Chao1 indices at day 0. By day 14, we observed a significant reduction in the Shannon index in the FOL-EAE group compared with either the U-EAE or the CO-EAE animals. On that day, the Chao1 index of the farnesol-treated mice was significantly lower than that of the corn-oil treated animals but was not different from the Choa1 index of the U-EAE group. On day 26, vehicle-treated mice showed an increased Shannon index when compared to untreated EAE mice, while no differences were detected between untreated and FOL-treated EAE mice or between vehicle-treated versus FOL-treated EAE mice. At this time point, the Chao1 alpha diversity measurement analysis did not show any significant differences among groups (Fig. 4B).

We next evaluated whether the treatment with FOL or vehicle would impact the intestinal microbiota’s overall composition at the species level (Fig. 4C). Our results indicate a treatment effect on the microbiota’s overall composition between EAE, vehicle-treated, and FOL-treated EAE mice. The changes in stool microbial composition from day 0 to day 26 are depicted in Supplemental figure 3. The data show that the disease’s progression in untreated mice resulted in significant alterations in the stool microbial structure at day 14, time of disease onset (Supplemental Fig. 3; p < 0.001 for day 0 vs. day 14). However, on average, the stool microbial structure of the EAE mice returned to pre-EAE induction composition at day 26 (p < 0.001 for day 14 vs. day 26; ns for day 26 vs. day 0). Progression of the disease resulted in a similar change in the stool microbial structure from day 0 to day 14 in both the FOL and the VEH groups (Supplemental Fig. 3B and 1C; p < 0.001). In contrast with the EAE group, however, these groups’ stool microbial structure never returned to pre-induction status.

We sought to determine if these differences were associated with changes in specific taxonomic units because of the observed group differences in stool microbiota composition. The treatment with vehicle and with FOL resulted in significant changes in Actinobacteria (day 14 and 26) and Verrucomicrobia (day 14) (Fig. 5A). FOL treatment increased the relative abundances of Actinobacteria compared with both untreated and vehicle-treated EAE mice at day 26. The increase in Actinobacteria after FOL treatment was significant at day 14 only when compared with untreated EAE mice, but not the vehicle-treated mice. The treatment with FOL increased Verrucomicrobia on day 14 versus untreated EAE and vehicle-treated mice. The phyla analysis also revealed significant differences in Firmicutes and Bacteroidetes; however, the differences were also observed on day 0, possibly due to cage-distribution variability (Supplemental Table 1; Supplemental Figure 3). Time exacerbated the differences in Bacteroidetes observed between treatments, with significantly higher abundances in FOL-treated versus untreated at day 0 (p < 0.05), at day 14 (p <0.01), and at day 26 (p < 0.001), FOL-treated versus vehicle-treated mice (day 0: ns; day 14: ns; day 26: p < 0.01). The treatment with the vehicle also increased the relative abundances of Bacteroidetes over time compared with no treatment, but only prior to the disease peak (day 0: ns; day 14: p < 0.001; day 26: ns). There was a significant group difference in Firmicutes abundance at day 0 and day 26, but not significant at day 14 (Fig. 5A). However, in a separate analysis, we compared the Firmicutes-to-Bacteroidetes (F/B) ratios among groups and timepoints (Fig. 5B). There was no significant difference among groups on days 0 and 14. Still, at day 26, the F/B ratios of the FOL-treated mice were on average significantly lower than those of the untreated (p < 0.01) and vehicle-treated (p < 0.05) mice (p < 0.05). There was no F/B ratio difference between untreated and vehicle-treated mice on that day (Fig. 5B).

Figure 5.

Oral treatment with FOL modifies the composition of the microbiota at the phylum level and changes the Firmicutes:Bacteroidetes (F/B) ratio. A. Analysis of the V4 region of the 16S rRNA at the phylum level at day 0 (disease induction), day 14 (disease onset), and day 26 (end of treatment) in FOL-EAE, CO-EAE, and U-EAE mice (number of samples analyzed per group and timepoint shown in figure 4). B. F/B ratios in the study groups. (individual values and mean ± 1SEM. Statistical analyses were performed as described in Methods: (A) 2-way ANOVA followed by Tukey’s multiple comparisons test (*, p < 0.05; **, p < 0.01; ***, p < 0.001), and (B) Kruskal-Wallis followed by Dunn’s multiple comparisons test (*, p < 0.05; **, p < 0.01).

We next compared the relative abundances at the family level using raw p values, as done at the phylum level (Fig. 6). When compared with untreated mice, FOL-treated mice had increased relative abundance in Bifodobacteriaceae (days 14 and 26), Erysipelotrichaceae (days 14 and 26), Ruminococcaceae (day 14), Clostridiaceae subfamily 1 (day 14), and Verrucomicrobioaceae (day 14). The abundance in Bacteroidales S24-7 was also enhanced in the FOL-treated mice compared to untreated EAE mice at days 14 and 26, but the difference already existed at day 0. The abundance of Lactobacillaceae followed a different pattern. It was significantly lower in the FOL-treated mice than in untreated mice at day 0, higher at day 14, and again lower at day 26. However, these differences mostly accounted for a significant decrease in the abundance of Lactobacillaceae of the untreated animals at day 14 compared with day 0 and a dramatic increase at day 26 (Fig. 6). The abundance in Lachnospiraceae was lower in the FOL-treated mice than in both the untreated and corn-oil-treated mice throughout the experiment, including day 0.

Figure 6.

Oral treatment with FOL modifies the relative abundances of members of the microbiota at the family level. Group comparisons are shown at day 0 (EAE induction; A), day 14 (disease onset; B), and day 26 (end of treatment; C). Individual data and group average (mean ± SEM) are shown. Group differences were evaluated using two-way ANOVAs followed by Tukey’s multiple comparisons test (*, p < 0.05; **, p < 0.01; ***, p < 0.001).

Corn oil also caused changes in the microbiota of EAE mice at the family level. These changes were similar to those observed in the FOL-EAE group for the Bacteroidales S24-7 group, Lactobacillaceae, Lachnospiraceae, Bifidobacteriaceae, and Erysipelotrichaceae. The comparison between corn-oil treated mice and FOL-treated animals showed the following significant differences (FOL-EAE vs. CO-EAE): 1) an increase in Bacteroidales S24-7 group (day 26), Bifidobacteriaceae (day 14), Erysipelotrichaceae (day 14), Ruminococcaceae (day 14), Clostridiaceae subfamily 1 (day 14) and Verrucomicrobioaceae (day 14), and 2) a decrease in Lachnospiraceae at days 14 and 26.

The analysis of the relative abundances at the genus level is shown in Supplemental figure 4. Treatment with FOL resulted in increased uncultured Bacteroidales S24-7 (days 0, 14, and 26), Lactobacillus (day 14), Bifidobacterium (days 14 and 26), Clostridium senso stricto 1 (day 14), Turicibacter (days 14 and 26), uncultured Erysipelotrichaceae, Faecalibaculum (day 14), and Akkermansia (day 14 and 26), versus untreated EAE mice. By contrast, Lachnospiraceae NK4A136 (days 0, 14, and 26), uncultured Lachnospiraceae (days 0, 14, and 26), and Prevotellaceae UCG-003 (day 14) were significantly reduced in FOL-treated mice when compared with untreated EAE mice.

4. Discussion

Our work demonstrates the protective effects of farnesol in murine EAE, a multiple sclerosis model used in academic and industry research to test new therapeutics. Our data show that daily oral treatment with 100 mg/kg FOL significantly attenuates the disease’s clinical severity. To our knowledge, the protective activity of FOL has not been reported in this model or MS. Still, it is consistent with previous reports showing neuroprotection in toxin-induced CNS inflammatory states [27,28]. In addition, we show that oral FOL administration significantly modifies the composition of the intestinal microbiome, promoting a reduction of the F/B ratio and increases in the relative abundances of Bifidobacteriaceae, Bacteroidales S24-7 group, and Erypelotrichaceae throughout the experiment, and Lactobacillaceae, Ruminococcaceae, Clostridiaceae 1, and Verrucomicrobioaceae at early stages of the disease. By contrast, FOL treatment reduced Lachnospiraceae and reduced Lactobacillaceae at later stages of the disease. Although preliminary, the microbiota findings provide evidence that FOL reshapes the dysbiotic microbiome of EAE mice and raise the possibility of a beneficial impact of FOL on the gut-brain axis as a mediator of its clinical activity.

The molecular and cellular mechanisms underlying the neuroprotective activity of FOL in experimental MS are unknown. However, several studies have reported that oxidative stress, autoimmune inflammation, and intraneuronal Ca²⁺ overload are involved in MS pathogenesis [42,43]. Thus, because FOL blocks neuronal Ca²⁺ channels, attenuates oxidative stress, and decreases pro-inflammatory cytokine secretion, it is tempting to propose that these mechanisms of action are responsible for the clinical benefits observed in experimental MS. Our study also shows that clinical improvement is associated with a significant decrease in spinal cord CD4⁺ T cells’ percentage. This finding suggests that FOL may improve the clinical presentation of EAE by yet another mechanism involving modulation of the autoimmune response, a response thought to be the most likely cause of MS in humans. Last, some studies have shown that HMG-CoA reductase inhibitors (statins) have immunomodulatory properties and reduce disease severity in EAE [44,45]. Others have reported that FOL also inhibits HMG-CoA-reductase [46,47]. It is thus possible that the disease protection conferred by FOL in our studies is mediated by the impact of the isoprenol on the mevalonate pathway. Such possibility will need to be explored in the future with investigations of the molecular consequences of FOL treatment on the mevalonate pathway in the CNS of EAE mice.

One potential limitation of our study is the confounding yet beneficial activity of corn oil, the vehicle for FOL. One recent work has reported that the oral administration of extra-virgin olive oil induced protection against EAE using a rat model [48]. The neuroprotection conferred by olive may be attributed to its high proportion of omega-3 fatty acids, described as promoting anti-inflammatory effects. In contrast, corn oil contains higher amounts of omega-6 fatty acids associated with pro-inflammatory effects. Thus, olive oil could have been a better vehicle for FOL than corn oil in our study. However, using LC-MS/MS, we found that that farnesol can be detected in olive oil but not in corn oil (data not shown), hence our choice of corn oil as vehicle. Our finding that corn oil provides significant neuroprotection in EAE mice (Fig. 1A) confirms previously published studies in which corn oil was used as vehicle [49] and highlights the importance of using clinically neutral vehicles in studies testing novel therapeutics in experimental MS, particularly in studies that focus on the microbiota. It is important to note here that the combination of FOL and corn oil provides superior clinical benefits compared to corn oil alone, raising the possibility that additive or synergistic mechanisms of action are at play. Here again, future studies will be needed to delineate the respective contribution of vehicle and FOL to neuroprotection, examine the potential for pharmacological interactions, and design FOL-containing “combo” drugs with enhanced therapeutic activity.

Interestingly, the spinal cords of EAE mice treated with FOL show significantly lower accumulations of monocyte/macrophages and dendritic cells (Fig. 2) and CD4⁺ T cells (Fig. 3A) than the spinal cords of vehicle-treated and untreated mice. The data further show a significant increase in the frequencies of Treg in the spinal cords of FOL-treated mice when compared to untreated EAE mice, suggesting that FOL treatment promotes Treg accumulation in the CNS (Fig. 3B). However, these differences, although statistically significant, are based on cell percentages, not the absolute number of cells infiltrated. Because of the small number of sorted cells, the effect of FOL on monocyte/macrophages, dendritic cells and Treg need to be interpreted with caution, and future studies with larger experimental groups allowing the pooling of multiple spinal cord samples will be needed to confirm our findings. We did not observe a significant increase in peripheral lymphoid organ Treg populations in FOL-treated mice (Supplemental Fig. 3), suggesting that EAE protection by FOL may not be associated with Tregs-mediated immunomodulation. Additional studies will thus be needed to further assess the role of Tregs in FOL-induced neuroprotection.

The stool microbiome analysis performed before and after disease onset offers a descriptive yet confirmatory evaluation of the temporal impact of EAE induction on the gut microbiota of a well-established model of MS [50,51]. Importantly, it provides intriguing insights on the effect of FOL on EAE intestinal dysbiosis and raises the possibility of participation of FOL-induced gut microbial changes in disease pathogenesis. We observed that FOL affected the overall composition of the microbiota and specifically the F/B ratio, with a significant decrease in the ratio at day 26 compared with vehicle-treated and untreated EAE mice (Fig. 5). The respective pathogenic or protective roles of Firmicutes and Bacteroides in diseases are uncertain. However, several studies have reported lower levels of Bacteroides (hence higher F/B ratio) in inflammatory disease of the gut [reviewed in [52]]. Other studies have shown that the F/B ratio increases in the stool isolated from patients with active MS compared with healthy individuals and is associated with an enhanced Th17 cell presence [53]. It is thus tempting to propose that in our model, the neuroprotective effect of FOL is mediated in part by a change in the composition of the gut microbiota, and subsequent attenuation of the autoimmune response, thus providing indirect support to gut-brain axis participation in MS pathogenesis. It remains to be elucidated whether the microbiome changes are the result of the direct impact of the treatment on the microbiome or the result of the reciprocal interactions that have been established between the microbiome and the immune system of those suffering from the disease.

In conclusion, our study provides pre-clinical evidence of the potential therapeutic benefit of farnesol in MS. The cellular and molecular mechanisms underlying neuroprotection have yet to be thoroughly investigated but could involve the previously reported activity of farnesol on voltage-gated Ca²⁺ channels, inflammatory cytokine secretion and oxidative stress, a decrease in monocyte CD11b⁺ cell subpopulations, including CD11c⁺ dendritic cells, and CD4⁺ T cells population with blunting of autoimmunity pathways, and an improvement of the gut microbiota composition (more Bacteroides, less Firmicutes). Studies are undergoing to characterize further the contribution of each of these mechanisms to the neuroprotective activity of the isoprenol and pave the way for translational research leading to novel treatment of patients with MS.

Supplementary Material

3

Supplemental figure 1. Oral treatment with FOL reduces the percentages of CD4⁺ T cells isolated from spinal cords during EAE. The percentage of CD4⁺ T cells in spinal cords was compared by flow cytometry at the end of the experiment (day 26) in untreated EAE mice (n = 4), vehicle-treated mice (n = 9), and FOL-treated mice (n = 8). A) Doublets were first discriminated for the identification of CD4-positively stained cells. B) The results of the analysis are depicted as mean +/− SEM of the percentages in singlets and total of the spinal cords and compared by Kruskal-Wallis analysis followed by Dunn’s multiple comparisons test (*, p < 0.05).

4

Supplemental figure 2. Oral treatment with FOL modifies the splenic monocyte/macrophage abundances at the peak of EAE but does not change splenic infiltration by CD4⁺ T cells or dendritic cells. Splenic cell analyses were performed by flow cytometry at day 26 (end of treatment) in U-EAE (n = 9), CO-EAE (n = 9), and FOL-EAE (n = 10) mice. A. Percentages of F4/80^neg granulocyte/monocytes; (far left), F4/80^int granulocyte/monocytes (left), monocyte-derived macrophages (MDM) (right), and monocyte-derived dendritic cells (far right) subpopulations in splenic population (individual data and mean ± 1 SEM). B. Percentages of CD4+ T cells in total splenic population (left), CD25⁺Foxp3⁺ cells in gated CD4⁺ T cells (middle), and CD25⁺Foxp3⁺CD4⁺ T cells in total splenic population (right).

5

Supplemental figure 3. Longitudinal study of the changes in the overall composition in EAE untreated and treated mice. PCoA analysis for the genus taxonomic level of OTUs identified in the analysis at days 0, 14, and 26 for untreated EAE mice (EAE), FOL-treated EAE mice, and vehicle-treated EAE mice. Data were analyzed using ADONIS (adjusted p values shown: *, p < 0.001).

6

Supplemental Table 1. PCoA analysis for the genus taxonomic level of OTUs identified in the analysis at day 0 for samples grouped by cages. Adjusted p values determined by ADONIS when comparing cages of mice at day 0

7

Supplemental figure 4. Oral treatment with FOL modifies the relative abundances of members of the microbiota at the genus level. Individual data and group averages (mean ± SD) are shown in stool samples collected at day 0 (disease induction; top), day 14 (disease onset; middle), and day 26 (end of treatment; bottom) in FOL-EAE, CO-EAE, and U-EAE mice (number of samples analyzed per group and timepoint shown in figure 4). Statistical analyses were performed using a two-way ANOVA followed by Tukey’s multiple comparisons test (*,p < 0.05; **,p < 0.01; ***, p < 0.001).