Modulating inflammatory monocytes with a unique microRNA gene signature ameliorates murine ALS

Abstract

Amyotrophic lateral sclerosis (ALS) is a progressive disease associated with neuronal cell death that is thought to involve aberrant immune responses. Here we investigated the role of innate immunity in a mouse model of ALS. We found that inflammatory monocytes were activated and that their progressive recruitment to the spinal cord, but not brain, correlated with neuronal loss. We also found a decrease in resident microglia in the spinal cord with disease progression. Prior to disease onset, splenic Ly6Chi monocytes expressed a polarized macrophage phenotype (M1 signature), which included increased levels of chemokine receptor CCR2. As disease onset neared, microglia expressed increased CCL2 and other chemotaxis-associated molecules, which led to the recruitment of monocytes to the CNS by spinal cord-derived microglia. Treatment with anti-Ly6C mAb modulated the Ly6Chi monocyte cytokine profile, reduced monocyte recruitment to the spinal cord, diminished neuronal loss, and extended survival. In humans with ALS, the analogous monocytes (CD14+CD16-) exhibited an ALS-specific microRNA inflammatory signature similar to that observed in the ALS mouse model, linking the animal model and the human disease. Thus, the profile of monocytes in ALS patients may serve as a biomarker for disease stage or progression. Our results suggest that recruitment of inflammatory monocytes plays an important role in disease progression and that modulation of these cells is a potential therapeutic approach.

Figure 2. Activation of the chemotaxis pathway in CD39⁺ resident microglia in the spinal cord but not the brain of SOD1 mice.

(A) Quantitative nCounter expression profiling of 179 inflammation-related genes was performed in spinal cord–derived CD39⁺ microglia from SOD1 mice and compared with non-Tg littermates at presymptomatic (60 days), onset (defined by body weight loss), and end stages. Heatmap shows genes with at least 2-fold-altered transcription levels. Each row of the heatmap represents an individual gene and each column an individual group in biological triplicate (n = 3 arrays for each group from pool of 4–5 mice at each time point). The relative abundance of transcripts is indicated by a color scale (red, high; green, median; blue, low). Bars show relative expression of significantly up- or downregulated genes in SOD1 mice from non-Tg littermates at each time point. Twenty significantly upregulated genes are shown in A, and 38 significantly downregulated genes are shown in B. Data for A and B represent mean ± SD. All shown genes were significantly affected (P < 0.05). The complete list of P values for each significantly affected gene is shown in Supplemental Table 1. Gene expression level was normalized against the geometric mean of 6 housekeeping genes (Cltc, Gapdh, Gusb, Hprt1, Pgk1, Tubb5). (C) Comparative analysis of significantly upregulated genes in CD39⁺ microglia from spinal cords of SOD1 mice at onset versus microglia isolated from the brain of the same mice.

Figure 1. Reciprocal expression of CD39 and Ly6C in CNS-resident microglia and inflammatory monocytes.

(A) Real-time quantitative RT-PCR (qRT-PCR) of Endtp1 (CD39) relative expression in adult microglia (CD11b⁺CD45^lo), alveolar macrophages (Alv Macs), neutrophils (Ly-6G⁺), T cells (CD3⁺), B cells (B220⁺), NK cells (NK1.1⁺), and regulatory T cells (CD4⁺CD25⁺Foxp3⁺) from naive adult C57BL/6J mice. (B) Endtp1 and (C) Ly6c1 (Ly6C) expression (qRT-PCR) in adult microglia and CD11b⁺ sorted cells from liver, heart, muscle, and kidney, and Ly6C^– and Ly6C⁺ monocyte subsets sorted with 6C3 mAb from PBMCs, spleen, and bone marrow. Expression levels were normalized to Gapdh. (D) Flow cytometry histograms show MFI of surface expression of CD39 and Ly6C in organ-specific CD11b-gated cells compared with isotype control (IC) (open histograms) from naive B6 mice. The numbers show percentage of CD39⁺ cells (left panels) and Ly6C^lo and Ly6C^hi monocytes (right panels). Each histogram panel represents a pool of 3–5 mice. Results are representative of 2 independent experiments.

Figure 3. Ly6C^hi monocytes in the spleen exhibit a proinflammatory profile 2 months prior to clinical disease onset and during disease progression in SOD1 mice.

Quantitative nCounter expression profiling of 179 inflammation-related genes showing significantly (A) upregulated and (B) downregulated genes in splenic Ly6C^hi monocytes compared with non-Tg littermates of the same mice analyzed in Figure 2 at presymptomatic (30 and 60 days of age), disease onset, and end stages. Data for A and B represent mean ± SD. All shown genes were significantly affected (P < 0.05). The complete list of P values for each significantly affected gene is shown in Supplemental Table 2. Gene expression level was normalized against the geometric mean of 6 housekeeping genes (Cltc, Gapdh, Gusb, Hprt1, Pgk1, Tubb5). (C) qRT-PCR analysis of Ahr mRNA expression in splenic Ly6C monocyte subsets sorted by flow cytometry from non-Tg, SOD1^WT, and SOD1^G93A mice at disease onset. Total RNA was isolated and pooled from 3–5 mice for each cell population. Expression levels were normalized to Gapdh. Data represent mean ± SEM. ***P < 0.001, Student’s t test (2-tailed). Results are representative of 2 independent experiments.

Figure 4. Changes in miRNA profiles of spleen-derived Ly6C^hi monocytes during the course of disease.

(A) miRNA profiling of splenic Ly6C^hi monocytes compared with non-Tg mice at presymtomatic (60 days), onset (defined by body weight loss), and end stages of disease was performed by rodent TLDA (containing 364 mouse miRNA assays; 2 arrays for each group, pool of 6–8 mice per group). Heatmap shows miRNAs with at least 2-fold-altered transcription levels. Microarray data were normalized using quantile (R software; http://www.r-project.org/) normalization. Each row of the heatmap represents an individual miRNA and each column an individual group in biological duplicate. The relative abundance of transcripts is indicated by a color scale (red, high; green, median; blue, low). (B) Summary of significantly affected miRNAs in splenic Ly6C^hi monocytes in SOD1 mice compared with non-Tg mice validated in Singleplex qRT-PCR. Data represent mean ± SD. All shown miRNAs were significantly affected (P < 0.05). miRNA expression level was normalized using ΔCt against U6 miRNA.

Figure 5. Top miRNA-mRNA interactions in splenic Ly6C^hi monocytes in SOD1 mice.

Ingenuity target filter analysis showing the top 10 miRNA-mRNA interactions based on identified affected mRNAs (Figure 3, A and B) and miRNAs (Figure 4, A and B) at disease onset.

Figure 6. Changes in miRNA profiles of spinal cord–derived CD39⁺ microglia during the course of disease.

(A) miRNA profiling of spinal cord–derived CD39⁺ microglia in SOD1 mice compared with non-Tg mice from the same mice as described in Figure 4. (B) Summary of significantly affected miRNAs in spinal cord microglia in SOD1 mice compared with non-Tg mice validated in Singleplex qRT-PCR. Validated miRNAs were selected based on their differential expression at all disease stages. Data represent mean ± SD. All shown miRNAs were significantly affected (P < 0.05). miRNA expression level was normalized using ΔCt against U6 miRNA.

Figure 7. Top miRNA-mRNA interactions in spinal cord CD39⁺ microglia in SOD1 mice.

Ingenuity target filter analysis showing the top 10 miRNA-mRNA interactions based on identified affected mRNAs (Figure 2, A and B) and miRNAs (Figure 6, A and B) at disease onset.

Figure 8. Ly6C^hi monocytes are recruited to the spinal cord with disease progression in SOD1 mice.

(A) FACS analysis of isolated spinal cord and (B) brain-derived mononuclear cells for CD11b, CD39, and Ly6C at 135 days in SOD1 mice. Numbers represent the percentage of CD11b-gated cells in each quadrant. (C) Proportional increase in inflammatory monocytes (black) and myeloid cells (gray) and decrease in CD39⁺ resident microglia (white) as related to total CD11b⁺ cells. Data represent mean ± SEM from 3 experiments (pool of 4–5 mice per group). (D) Expansion of Ly6C monocytes from Ly6C^lo to Ly6C^hi during disease progression in the spinal cord. Numbers represent the percentage of cells in each quadrant. (E) Ly6C expression is increased during disease progression on recruited monocytes but not on resident microglia. The numbers show percentage of Ly6C^lo (left) and Ly6C^hi (right) monocytes. Open profiles represent staining pattern with an IC antibody; solid red profiles indicate CD11b⁺CD39⁺ microglia; and green profiles show recruited CD11b⁺Ly6C⁺ monocytes. Each panel represents a pool of 5 mice. Results are representative of 3 independent experiments. (F) qRT-PCR analysis of Ccr2 and Ccl2 mRNA expression in FACS-sorted CD39⁺ microglia and Ly6C^hi monocytes from spinal cords of SOD1^WT and SOD1^G93A mice. Total RNA was isolated and pooled from 5 mice for each cell population. Expression levels were normalized to Gapdh. Data represent mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, 1-way ANOVA followed by Dunnett’s multiple-comparison post hoc test. Results are representative of 2 independent experiments.

Figure 9. SOD1 spinal cord microglia induce recruitment of Ly6C⁺ monocytes.

(A) FACS analysis of myeloid cells isolated from recipient WT and SOD1 brains at disease onset 48 hours after transplantation (intracranial) with spinal cord–derived CD39⁺ microglia (5 × 10⁴ cells) from donor WT and SOD1 mice at disease onset. Cells were gated using Annexin V and 7-AAD to eliminate apoptotic and necrotic cells. Numbers represent the percentage of cells in each quadrant. Panels are representative of 3–4 mice in each group. (B) Quantitative analysis showing absolute numbers of recruited Ly6C⁺ cells per hemisphere of injected mice. Data represent mean ± SEM (3–4 mice per group). **P < 0.01, ***P < 0.001, 2-way ANOVA with Bonferroni post hoc test.

Figure 10. Reciprocal expression of CD39 and Ly6C in CNS-resident microglia and bone marrow–derived monocytes in SOD1 chimeric mice.

SOD1^G93A and non-Tg littermates were transplanted with syngeneic bone marrow cells from CX3CR1^GFP/+ mice. Spinal cords were analyzed at presymptomatic (60 days), disease onset, and end stages. (A) GFP⁺ recruited IBA1⁺ monocytes in lumbar spinal cord of non-Tg– and SOD1-Cx3cr1^GFP+/– mice. Scale bar: 500 μm. (B) Confocal images of GFP⁺ recruited monocytes (IBA1⁺GFP⁺; white arrowheads) and resident microglia (IBA1⁺GFP^–; yellow arrowhead) in ventral horns of non-Tg– and SOD1-Cx3cr1 chimeric mice at 120 days of age. Representative confocal images (5–6 mice per group). Scale bars: 50 μm. (C) Quantitative analysis showing the kinetics of bone marrow–derived CX3CR1^GFP/+ monocytes recruited into spinal cords during disease at 90, 120, and 145 days of age in SOD1-Cx3cr1^GFP+/– chimeric mice. Data represent mean ± SEM (5–6 mice per group). ***P < 0.001, 1-way ANOVA followed by Dunnett’s multiple-comparison post hoc test. (D) FACS analysis of CD39 and Ly6C expression in spinal cord–derived populations of microglia (MG) and peripheral monocytes (PMs) isolated from non-Tg– and SOD1-chimera mice at 120 days of age. Note: CD11b⁺GFP⁺ gated peripheral monocytes express Ly6C^hi and do not express CD39, whereas all resident microglia express CD39 and are negative for Ly6C^hi. (E) Expansion of the recruited Ly6C^hiCX3CR-GFP^lo monocyte subset in the spinal cord of SOD1 mice during disease progression. Numbers in D and E represent the percentage of CD11b-gated cells in each quadrant. Each panel represents a pool of 4–5 mice.

Figure 11. Ly6C^hi monocytes proliferate and CD39⁺ microglia undergo apoptosis during disease progression in the spinal cord of SOD1 mice.

Spinal cord–isolated myeloid cells at onset (90 days), early symptomatic (120 days), and late symptomatic (135 days) stages from SOD1^WT and SOD1 mice were analyzed. (A) Microglia viability was evaluated using Annexin V and 7-AAD for apoptotic and necrotic cells, respectively. No significant apoptosis was detected in Ly6C⁺ monocytes (data not shown). Numbers represent the percentage of cells in each quadrant. (B) Quantification of microglia viability reveals an approximately 2.5-fold increase in microglial apoptosis at 90, 120, and 135 days in comparison to wild-type microglia. Data represent mean ± SEM. *P < 0.05, **P < 0.01. (C) Proliferation of CD39⁺ resident microglia and Ly6C⁺ monocytes assessed by BrdU incorporation. BrdU was injected (i.p.) daily for 5 consecutive days before the spinal cords were analyzed. Wild-type mice received the same course of BrdU injection. Spinal cords were excised 5 days after the first BrdU injection. G₁-gated CD11b⁺CD39⁺ microglia; G₂-gated Ly6C^hi; and G₃-gated Ly6C^lo monocytes. Flow cytometric analysis was based on live cell population after exclusion of Annexin V–and 7-AAD–positive cells. Numbers represent the percentage of cells in each quadrant. (D) Ly6C^hi monocytes proliferate 3- to 4-fold more than Ly6C^lo cells during the disease course. Data represent mean ± SEM (pool of 3–4 mice per group). ***P < 0.001, Student’s t test (2-tailed).

Figure 12. Anti-Ly6C mAb treatment delays disease onset and extends survival in SOD1 mice.

SOD1^G93A mice were treated i.p. with IgG2a (IC; 100 μg, n = 11) or 100 μg anti-6C3 mAb (n = 11) every other day starting at the onset of the disease. (A) Kaplan-Meier analysis of the probability of surviving of SOD1 as function of age. Mantel-Cox’s F-test comparison showed groups treated with 100 μg IC versus anti-Ly6C (P = 0.0097). (B) Time-to-event analysis for disease neurologic onset (neurological severity score of 2). Disease onset was significantly delayed (P = 0.0127) by anti-Ly6C (100 μg) treatment. (C) Rotarod performance of IC- and anti-Ly6C–treated groups as a function of age. Data represent mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, IC compared with anti-Ly6C groups by factorial ANOVA and Fisher’s least significant difference post hoc test. (D) Weight loss plotted for IC- and anti-Ly6C–treated groups. Data represent mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, 2-way ANOVA, Bonferroni post hoc test. (E) Duration of an early disease phase (from onset to 5% weight loss) and a later disease phase (from 5% weight loss to end stage). Data represent mean ± SEM. *P < 0.05, ***P < 0.001, 1-way ANOVA.

Figure 13. Anti-Ly6C treatment affects the phenotype of splenic Ly6C^hi monocytes in SOD1 mice.

(A) SOD1 mice were treated with systemic (i.p.) injection every other day with IgG2a IC (100 μg; n = 5) or anti-Ly6C mAb (100 μg; n = 6) starting at disease onset (body weight loss). After 1 month of treatment (120 days of age), the mice were sacrificed and their splenic CD11b⁺Ly6C^hi sorted cells were analyzed with quantitative nCounter profiling for 179 inflammation-related genes. Bars show affected genes with at least 2-fold-altered transcription levels. Gene expression levels were normalized against the geometric mean of 6 housekeeping genes (Cltc, Gapdh, Gusb, Hprt1, Pgk1, Tubb5). (B) Validation of Il1b, Il6, Tnf, and Tgfb1 by qRT-PCR (TaqMan) in spleen- and spinal cord–derived Ly6C^hi monocytes. Expression levels were normalized to Gapdh. Results represent data from 2 independent experiments, each with 3–6 mice. Error bars represent mean ± SEM. **P < 0.01, ***P < 0.001, Student’s t test (2-tailed). (C) Expression of Il1b, Il6, Tnf, and Tgfb1 (qRT-PCR) in spleen- and spinal cord (SC)–derived Ly6C^hi monocytes in untreated WT and SOD1 mice at 120 days of age. Expression levels were normalized to Gapdh. Data represent mean ± SEM from 6 mice. *P < 0.05, ***P < 0.001, 1-way ANOVA followed by Dunnett’s multiple-comparison post hoc test.

Figure 14. Anti-Ly6C mAb treatment decreases infiltration of Ly6C^hi monocytes into the spinal cord and attenuates neuronal loss.

SOD1 mice were treated as described in Figure 12. (A) FACS analysis of Ly6C⁺ monocytes in the spinal cord of anti-Ly6C–treated SOD1 mice compared with the IC group 30 days after treatment. Cells were gated using Annexin V and 7-AAD to eliminate apoptotic and necrotic cells. Numbers represent the percentage of CD11b-gated cells in each respective quadrant as indicated. Pooled data from 5 mice are shown. (B) Significantly reduced proportion of Ly6C⁺ monocytes and increased numbers of CD39⁺ microglia among CD11b⁺ cells 50 days after α-Ly6C treatment. (C) Significant reduction in CD11b⁺CD169⁺ monocytes was detected after 50 days of α-Ly6C treatment. Numbers represent the percentage of cells in each gate. (D) Representative confocal images stained for NeuN (green; neurons), IBA1 (blue; myeloid cells), and CD169 (red; recruited monocytes) of whole mount lumbar axial sections of spinal cords from IC- and Ly6C-treated mice at the end stage (140 days old). Scale bar: 500 μm. Boxed areas showed insets at high magnification. Scale bars: 200 μm. (E) Quantitation of neurons (NeuN⁺) and recruited monocytes (IBA1⁺/CD169⁺) in ventral and dorsal horns in the spinal cord of SOD1 mice treated with isotype control or anti-Ly6C mAbs (n = 6–8 per group). Results are representative of 2 independent experiments. Error bars represent mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t test (2-tailed).

Figure 15. miRNA profiling of peripheral monocytes (CD14⁺CD16^–) in ALS patients demonstrates an inflammatory phenotype analogous to that observed in SOD1 mice.

(A) Quantitative nCounter expression profiling of blood-sorted CD14⁺CD16^– monocytes for 664 miRNAs in sporadic ALS (n = 8) and relapsing-remitting MS patients (n = 8) versus healthy controls (HC; n = 8). Heatmap of uncentered Pearson correlation was used as the distance metric with average linkage for the unsupervised hierarchical clustering. P < 0.01, nonparametric Kruskal-Wallis test, significance based on Benjamini-Hochberg FDR; selected FDR limit, 0.05. (B) Bars show fold differences of significantly affected miRNAs in ALS and MS subjects versus healthy controls. miRNA expression level was normalized against the geometric mean of 5 internal housekeeping genes (ACTB, B2M, GAPDH, RPL19, RPLP0). Data represent mean ± SD. All shown miRNAs were significantly affected (P < 0.05). The complete list of P values for each significantly affected miRNA is shown in Supplemental Table 4.

Figure 16. Identification of a unique microRNA signature in CD14⁺CD16^– blood monocytes from ALS subjects.

(A) PCA of the identified affected miRNAs between ALS and MS subjects versus healthy controls with spatial miRNA distribution. (B) qRT-PCR validation of 6 selected miRNAs in an independent healthy control and sALS cohort. Relative expression in HCs, MS, sALS, and fALS was calculated using the comparative Ct (2^–ΔΔCt) method. miRNA expression level was normalized against U6 miRNA. PCRs were run in duplicate per subject. Each data point represents an individual subject. Horizontal bars denote mean of miRNA expression for each group. **P < 0.01, ***P < 0.001, 1-way ANOVA followed by Dunnett’s multiple-comparison post hoc test.

Figure 17. Immune gene profiling of peripheral monocytes (CD14⁺CD16^–) in ALS patients.

Quantitative nCounter expression profiling of CD14⁺CD16^– monocytes for 511 immune- and 184 inflammation-related genes in sALS (n = 10) and fALS-SOD1 patients (n = 4) versus healthy controls (n = 10). (A) Heatmap of unsupervised hierarchical clustering (Pearson correlation) shows significantly affected genes (P < 0.01, nonparametric Kruskal-Wallis test, significance based on FDR Benjamini-Hochberg; selected FDR limit, 0.05). (B) Bars show fold differences of significantly affected genes in sALS and fALS subjects versus healthy controls. Gene expression level was normalized against the geometric mean of 15 internal-reference housekeeping genes (ABCF1, ALAS1, EEF1G, G6PD, GAPDH, GUSB, HPRT1, OAZ1, POLR1B, POLR2A, PPIA, RPL19, SDHA, TBP, TUBB). Data represent mean ± SD. All shown miRNAs were significantly affected (P < 0.05). The complete list of P values for each significantly affected miRNA is shown in Supplemental Table 5.

Figure 18. Identification of a unique gene signature in CD14⁺CD16^– blood monocytes from ALS subjects.

(A) PCA analysis of the identified affected genes between sALS and fALS subjects versus healthy controls with spatial gene distribution. (B) qRT-PCR validation of 8 selected genes in an independent cohort that were the most significantly upregulated or downregulated. Relative expression in sALS and fALS against healthy controls were calculated using the 2^–ΔΔCt method. Gene expression level was normalized against the geometric mean of 3 housekeeping genes (GAPDH, TUBB, and GRB2). PCRs were run in duplicate per subject. Each data point represents an individual subject. Horizontal bars denote mean of gene expression for each group. **P < 0.01, ***P < 0.001, 1-way ANOVA followed by Dunnett’s multiple-comparison post hoc test.

Figure 19. Top miRNA-mRNA interactions in CD14⁺CD16^– blood monocytes from ALS subjects.

Ingenuity target filter analysis showing miRNA-mRNA interactions based on identified significantly affected miRNAs (Figure 15) and mRNAs (Figure 17) in ALS subjects.