Myeloid and lymphoid expression of C9orf72 regulates IL-17A signaling in mice

Abstract

A mutation in C9ORF72 is the most common cause of Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Dementia (FTD). Patients with ALS or FTD often develop autoimmunity and inflammation that precedes or coincides with the onset of neurological symptoms, but the underlying mechanisms are poorly understood. Here, we knocked out murine C9orf72 in seven hematopoietic progenitor compartments by conditional mutagenesis and found that myeloid lineage C9orf72 prevents splenomegaly, loss of tolerance, and premature mortality. Furthermore, we demonstrated that C9orf72 plays a synergistic role in lymphoid cells to prevent interleukin (IL)-17A production and neutrophilia. Mass cytometry identified early and sustained elevation of the co-stimulatory molecule CD80 expressed on C9orf72-deficient mouse macrophages, monocytes, and microglia. Enrichment of CD80 was similarly observed in human spinal cord microglia from patients with C9ORF72-mediated ALS compared with non-ALS controls. Single cell RNA sequencing of murine spinal cord, brain cortex, and spleen demonstrated coordinated induction of gene modules related to antigen processing and presentation and antiviral immunity in C9orf72-deficient endothelial cells, microglia, and macrophages. Mechanistically, C9ORF72 repressed the trafficking of CD80 to the cell surface in response to toll-like receptor agonists, interferon-γ, and IL-17A. Deletion of Il17a in C9orf72-deficient mice prevented CD80 enrichment in the spinal cord, reduced neutrophilia, and reduced gut T helper type 17 cells. Lastly, systemic delivery of an IL-17A neutralizing antibody augmented motor performance and suppressed neuroinflammation in C9orf72-deficient mice. Altogether, we show that C9orf72 orchestrates myeloid co-stimulatory potency and provides support for IL-17A as a therapeutic target in ALS/FTD.

One Sentence Summary:

The ALS-associated C9orf72 gene product opposes IL-17A-dependent inflammation in myeloid and lymphoid cells.

Editor’s Summary:

C9ORF72 is commonly mutated in Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Dementia (FTD) and loss of function has been associated with autoimmune inflammation. Limone et al. find that hematopoietic loss of C9orf72 expression drives excess IL-17A inflammation, whereas loss of C9orf72 in myeloid cells is sufficient to cause severe autoimmunity. C9orf72 deficient mice had more myeloid cells with high surface expression of co-stimulatory molecule CD80 which was potentiated by IL-17A. Patients with C9ORF72-related ALS similarly showed enrichment of CD80 in spinal cord microglia. IL-17A neutralizing antibody therapy in C9orf72 deficient mice reduced neuroinflammation and support further investigation of IL-17A-based therapies for ALS or FTD. –Molly Ogle

INTRODUCTION

The most common inherited cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) is a hexanucleotide (GGGGCC) repeat expansion within the first intron of Chromosome 9 Open Reading Frame 72 (C9ORF72). The mutation accounts for 20-40% of familial ALS cases and 2-8% of sporadic ALS cases and is enriched in people of European descent (1). The mutation is transcribed into long repetitive ribonucleic acid (RNA) species and undergoes repeat associated non-AUG translation into dipeptides, both of which can produce toxic gain-of-function (GOF) effects when overexpressed in the central nervous system or in cultured neurons (2–4). Independent of these gain-of-function (GOF) effects, epigenetic repression and failure of RNA polymerase to read through the mutation in brain and hematopoietic cells leads to a loss-of-function (LOF) mechanism and reduction of the gene normally encoded at the C9ORF72 locus (5). C9ORF72 exhibits GTPase activating protein (GAP) activity towards small GTPases of the RAB and ARF family that function as molecular switches to mediate vesicle trafficking, actin cytoskeletal rearrangement, and autophagic flux (6–9). In mice, the C9ORF72 ortholog (C9orf72) suppresses age-related inflammation and autoimmunity, as well as promoting phagolysosomal clearance of dipeptides to restrain GOF toxicities of a transgene-encoded C9ORF72 repeat expansion (10–15). A key question is whether and how C9ORF72 LOF-driven peripheral autoimmunity imparts risk of motor neuron disease.

Chronic inflammation may be a contributor to ALS (16), as autoimmune diseases often precede an ALS/FTD diagnosis (17, 18) and neuroinflammation at disease onset correlates with the rate of motor decline (19, 20). The ratio of circulating T helper type 17 (Th17) to T regulatory cells (T_reg) cells is elevated in patients with ALS who experience a rapid disease course (21) which supports a role for adaptive immune cells in disease progression. Interleukin (IL)-17A , a key proinflammatory cytokine produced by Th17 cells and elevated in many auto-inflammatory disorders (22), is elevated in ALS/FTD plasma and cerebral spinal fluid, including in patients with a C9ORF72 mutation (23–26). However, the extent to which GOF or LOF sequelae of the C9ORF72 mutation contribute to inflammation, autoimmunity, and neural degeneration in patients with ALS and FTD, as well as the relative importance of the IL-17A inflammatory axis, remains incompletely understood.

Therapeutic efforts to restore C9ORF72 function in ALS/FTD patients will benefit from the knowledge of cell types and pathways in which this gene acts to promote organismal health. However, C9ORF72 cellular function remains poorly understood, in part because the gene is widely expressed in all three germ layers. The profound inflammatory response that manifests in C9orf72 LOF mice (10–13) has prompted investigation into whether this gene product serves an important function in the immune system. Reciprocal bone marrow transplantations established that C9orf72 is required in both radio-sensitive and radio-resistant cells to oppose systemic inflammation and fatal autoimmunity (11). Studies in which C9orf72-deficient mice that were reared in different environments, treated with antibiotics, or received fecal transplant showed that neuroinflammation and fatal autoimmune disease occur in response to signals derived from gut bacteria (27). C9orf72 acts in macrophages to suppress the release of inflammatory cytokines when exposed to gut bacteria (27) implicating the innate immune response as a governor of organismal disease severity. Furthermore, selective depletion of C9orf72 within Cx3cr1- or LysM-expressing cells in mice caused inflammation that was dependent on the cytosolic DNA sensor Stimulator of interferon genes (STING) (28). However, because spontaneous autoimmune disease was not reported in C9orf72-deficient or myeloid conditional mutant mice reared in that environment (28), it has remained unclear whether the function of C9orf72 in myeloid cells is required to maintain immune tolerance and to what extent, if any, C9orf72 functions in other hematopoietic lineages to oppose systemic and nervous system inflammation.

Here, we have addressed this knowledge gap by performing a conditional mutagenesis screen of C9orf72 activity across discrete hematopoietic lineages and by profiling gene expression and protein abundance of C9orf72 LOF mice with single-cell resolution. Targeted mutagenesis confirmed a general role for C9orf72 in the hematopoietic system and a specific role in myeloid cells and lymphoid cells to oppose systemic inflammation and autoimmunity. We found that C9orf72-deficient microglia and endothelial cells participate in a coordinated induction of neuroinflammation that is characterized by dynamic activation of antigen processing and presentation, cytokine production, and antiviral immunity. Further delineating the function of C9ORF72, we identify Il-17a-dependent accumulation of the potent co-stimulatory molecule CD80 on the surface of C9orf72-deficient macrophages, microglia, and brain-infiltrating monocytes and demonstrate therapeutic improvement of motor function and neuroinflammation with IL-17A neutralization. Collectively, our findings identify roles for C9orf72 in both lymphoid and myeloid lineages, where it confers protection against pathogenic Th17-associated inflammation and maintains neural health.

RESULTS

Myeloid lineage-expressed C9orf72 restrains autoimmune inflammation and supports a normal lifespan.

C9ORF72 and its murine ortholog are enriched in monocytes, granulocytes, erythrocytes, dendritic cells, B cells, and activated T cells (fig. S1, A and B) (29). Our previous bone marrow transplant studies suggested a potential role for C9orf72 in the hematopoietic system to oppose fatal autoimmunity (11). To test the hypothesis that C9orf72 promotes organismal health by acting in divergent blood lineages, we crossed mice carrying two copies of a floxed C9orf72 (C9orf72^2Lox/2Lox) allele (30) with seven different Cre recombinase lines to generate cohorts of conditional knockout mice (C9orf72^2Lox/2Lox Cre+) and littermate controls (C9orf72^2Lox/2Lox Cre−) for study (Fig. 1, A and B and fig. S1C). Vav guanine nucleotide exchange factor 1 (Vav1)-Cre and poly(I:C) induced MX dynamin-like GTPase 1 (Mx1)-Cre serve as positive controls for the depletion of C9orf72 within all hematopoietic cells (31, 32). Lysozyme 2 (LysM)-Cre was used to inactivate C9orf72 broadly in the myeloid lineage (33), whereas CD2-Cre was chosen to deplete C9orf72 in common lymphoid progenitors (34). Additionally, we targeted C9orf72 for depletion within CD19⁺ B cells (CD19-Cre) (35), CD4⁺ and CD8⁺ T cells (CD4-Cre) (36), and Forkhead box P3 (FoxP3⁺) T_reg cells (Foxp3-Cre) (Fig. 1B) (37). For each intercross, the presence of Cre recombinase was associated with a reduction in the size of the C9orf72 locus (fig. S1D). C9orf72 expression was completely abrogated in peripheral blood from C9orf72^2Lox/2Lox;Vav1-Cre+ mice (fig. S1E) and C9orf72 was selectively depleted in sorted CD11c⁺ dendritic cells (DCs) and F4/80⁺ macrophages, but not B cells or neutrophils, in C9orf72^2Lox/2Lox;LysM-Cre+ mice (fig. S1F and table S1) as expected.

Fig. 1. C9orf72 in myeloid and lymphoid lineages opposes autoimmunity and IL-17A production.

(A) C9orf72 gene locus schematic and conditional mutagenesis strategy. Wt: C9orf72 endogenous allele; 2Lox (2L): LoxP sites flanking exon 4-5; Δ: Cre-recombined allele with premature stop codon. (B) Hematopoietic stem and progenitor populations targeted for Cre recombination (Vav1: hematopoietic; Mx1: myeloid & lymphoid; LysM: myeloid; CD2: lymphoid; CD4: CD4⁺ and CD8⁺ T cells; CD19: B cells; Foxp3: T_reg cells). (C) Spleen weight of the following genotypes: 5-month-old Vav1-Cre− (n=7 per group) or Vav1-Cre+ (n=6 per group); 4-month-post-poly(I:C) Mx1-Cre− (n=10 per group) or Mx1-Cre+ (n=9 per group); 9-month-old LysM-Cre− (n=10 per group) or LysM-Cre+ (n=10 per group); 12-month-old CD2-Cre− (n=6 per group) or CD2-Cre+ (n=5 per group); 11-month-old, CD19-Cre− (n=8 per group) or CD19-Cre+ (n=8 per group); 15-month-old, CD4-Cre− (n=8 per group), CD4-Cre+ (n=5 per group); 11-month-old, Foxp3-Cre− (n=10 per group), Foxp3-Cre+ (n=8 per group); Each dot represents one mouse (Cre− controls filled, Cre+ open); data were analyzed by unpaired Student’s t-test (Vav1, LysM, CD2, CD4, Foxp3) or Mann-Whitney test (Mx1, CD19) between Cre+ and Cre− mice within each strain. (D) Spleens from mice with conditional deletion of C9orf72 in the indicated lineages. (E) Survival of mice reared at Harvard University facility during the same two-year period. Animals that met pre-defined euthanasia criteria were considered a lethal event. C9orf72^+/+ and C9orf72^−/− mice euthanized for reasons unrelated to the pre-defined criteria were censored at time of euthanasia; data were analyzed by Gehan-Breslow-Wilcoxon test. (F to I) Peripheral blood analysis of male and female mice evaluated in the following animals: 5-month-old Vav1-Cre− (n=4-9 per group) or Vav1-Cre+ (n=6-11 per group); 4-month post-poly(I:C) Mx1-Cre− or Mx1-Cre+ (n=9 per group); 9-month-old LysM-Cre− (n=6-9 per group) or LysM-Cre+ (n=7-15 per group); 12-month-old CD2-Cre− (n=5-8 per group) or CD2-Cre+ (n=4-10 per group); 11-month-old, CD19-Cre− (n=8-11 per group) or CD19-Cre+ (n=6-8 per group); 15-month-old, CD4-Cre− (n=8 per group), CD4-Cre+ (n=5 per group); 11-month-old, Foxp3-Cre− (n=4-10 per group), Foxp3-Cre+ (n=3-8 per group). Cre− (filled dot) and Cre+ (unfilled dot) mice are shown for each genotype. Plasma IgM and IgG autoantibodies against 124 self-antigens (F); data were analyzed by Mann-Whitney test. Blood platelet counts (G); data were analyzed by Student t-test (LysM, CD2, CD19, Foxp3) or Mann-Whitney test (Vav1, CD4). Blood neutrophil counts (H); data were analyzed by Student t-test (Vav1, LysM, CD2, CD19, Foxp3) or Mann-Whitney test (CD4). (I) Heatmap of individual plasma cytokine concentrations; arrowhead points to IL-17A, color normalized from low (blue) to high (red); data were analyzed by two-way ANOVA with Sidak’s correction for multiple comparisons. (J) Summary of phenotypes in C9orf72 conditional mutant mice. For all panels: * p < 0.05, ** p < 0.01, ns: not significant.

Complete knockout of C9orf72 in mice (C9orf72^−/−) is characterized by splenomegaly as an early and sensitive indicator of inflammatory disease progression (10–15). Hematopoietic cell and myeloid lineage-specific conditional knockouts (C9orf72^2Lox/2Lox;Vav1-Cre+, Mx1-Cre+, and LysM-Cre+ mice) exhibited splenomegaly relative to littermate Cre− controls, but spleen weight was not altered in the lymphoid-specific Cre+ lines studied (Fig. 1, C and D). C9orf72^−/− mice experience premature death (Fig. 1E, red line), therefore, to test whether myeloid lineage conditional deletion of C9orf72 was sufficient to cause premature death, we aged a cohort of C9orf72^2Lox/2Lox;LysM-Cre+ (n=16) and LysM-Cre− control littermates (n=9) to 600 days and compared their survival to other C9orf72^+/+ (n=869) and C9orf72^−/− (n=838) mice that were reared contemporaneously (Fig. 1E). All control C9orf72^2Lox/2Lox;LysM-Cre− mice survived beyond 500 days of age (781-day median survival) and their survival did not differ from C9orf72^+/+ controls (Fig. 1E). C9orf72^2Lox/2Lox;LysM-Cre+ animals (547-day median survival) died significantly faster than LysM-Cre− controls (p = 0.0252) (Fig. 1E), whereas C9orf72^2Lox/2Lox;LysM-Cre+ mice lived significantly longer than C9orf72^−/− mice (462-day median survival) (p = 0.0481) (Fig. 1E). Male C9orf72^−/− (439-day median survival) tended to have a shorter median survival time than female C9orf72^−/− mice (499-day median survival) (p < 0.0001), but the C9orf72^2Lox/2Lox;LysM-Cre+ cohort was not sufficiently powered to assess differences between sexes (fig S1G). The C9orf72^2Lox/2Lox;LysM-Cre+ animals exhibited wasting, hepatomegaly, and labored breathing, which were also commonly displayed by C9orf72^+/− and C9orf72^−/− mice (11). These results indicate that myeloid C9orf72 prevents splenomegaly, a feature of systemic inflammation, and promotes a longer lifespan.

C9orf72 acts in diverse blood lineages to prevent autoimmunity.

Because depletion of C9orf72 in the myeloid lineage (LysM-Cre) caused premature mortality with delayed kinetics relative to C9orf72^−/− mice, we reasoned that our conditional mutagenesis survey could yield insight into the epistasis of inflammatory and autoimmune phenotypes that develop upon loss of C9orf72 (11). Consistent with a necessary role for C9orf72 in hematopoietic cells, C9orf72^2Lox/2Lox;Vav1-Cre+ animals recapitulated the autoinflammatory phenotype of C9orf72^−/− mice. Vav1-Cre+ mice had elevated plasma IgM and IgG autoantibodies (Fig. 1F and fig. S2, A and B), decreased circulating platelets indicative of pseudothrombocytopenia, a phenomena caused by autoantibodies that aggregate platelets, (Fig. 1G) (27) and neutrophilia (Fig. 1H). Cytokines and chemokines significantly (p < 0.05) elevated in C9orf72^2Lox/2Lox;Vav1-Cre+ plasma included IL-17A, IL-23, IL-6, IL-22, IL-15/15R, IL-3, IL-31, CXCL1 (GRO-α), CCL2 (Mcp1), and IL-28 (Fig. 1I). Conditional depletion of C9orf72 within either the myeloid (LysM-Cre+) or lymphoid (CD2-Cre+) lineage was sufficient to cause autoimmune phenotypes characterized by increased autoantibodies in plasma (Fig. 1, F and G), but did not lead to systemic elevation of cytokines or neutrophilia (Fig. 1, H and I). Reduction of C9orf72 in B cells or T cells alone did not cause detectable immune phenotypes (Fig. 1, C to I). Although IgG autoantibodies did not differ appreciably upon myeloid C9orf72 depletion, the IgM autoantibodies that were elevated in C9orf72^2Lox/2Lox;LysM-Cre+ and C9orf72^2Lox/2Lox;Mx1-Cre+ mice targeted proteins important for blood brain barrier integrity (Aquaporin 4, Vitronectin, Elastin, Collagen IV) (38–40) and microglia- and astrocyte-mediated synaptic pruning by the complement cascade (C9, C3, C4, C1q, Factor H) (fig. S2A) (41). Taken together, our results indicate that C9orf72 functions separately in myeloid and lymphoid lineages to promote immunological tolerance and cooperatively in these lineages to prevent systemic elevation of IL-17-related inflammatory cytokines and neutrophils (Fig.1J).

C9orf72 deficiency alters immune cell surface effector protein abundance.

We next sought to identify C9orf72-regulated factors that underlie pathogenesis in this model. We reasoned that deep immunophenotyping of the spleen might yield mechanistic insight because resident professional antigen-presenting cells (APCs) such as macrophages and DCs integrate signals from the environment, display foreign and self-antigens on their cell surface, and provide signals that govern the activity of T cells (42). A panel of 34 heavy isotope-labeled antibodies was designed that included 16 antigens to assign cellular identity and 18 functional effectors with a focus on interactions between APCs and T cells (Fig. 2A and table S2).

Fig. 2. Macrophage-expressed C9orf72 restrains CD80 co-stimulation.

(A) Classification of antibody targets in mass cytometry panel. (B) Mass cytometry workflow from spleen of male and female littermates at 2- or 8-months of age. (C) Gating scheme of 9 defined splenocyte populations. (D) Cellular abundance of each gated population from spleen. Dendritic cell (DC). 2mo: C9orf72^+/+ (n=7 per group); C9orf72^−/− (n=9 per group); 8mo: C9orf72^+/+ (n=7 per group); C9orf72^−/− (n=8 per group), dots represent individual mice; data were analyzed by one-way ANOVA with Sidak’s correction for multiple comparisons (Macrophage, CD8⁻ DC) or Kruskal-Wallis test with Dunn’s correction for multiple comparisons (CD8⁺ DC, CD4⁺CD8⁺T, Monocyte, Neutrophil, CD4⁺T, CD8⁺T, B cell). (E) Average effector median staining intensity on surface of gated splenic populations. For each effector, color scale was normalized from lowest (blue) to highest (red) expression across populations. Data were analyzed by Student’s t-test or Mann-Whitney test with Bonferroni’s correction for multiple comparisons, * p < 0.0056. (F) Representative flow cytometry plots of CD11b and CD80 among CD45⁺ cells in the indicated tissues from C9orf72^+/+ (top) or C9orf72^−/− (bottom). mLN, mesenteric lymph node. (G) Quantification of CD11b^hi CD80⁺ cells in the indicated tissues; dots represent individual mice; data analyzed by two-way ANOVA with Sidak’s correction for multiple comparisons. (H) Percentage of OTII T cell receptor transgenic T cells with diluted carboxy fluorescein succinimidyl ester (CFSE) signal after 4 days in co-culture with tissue enriched Ovalbumin (OVA) peptide-loaded macrophages; dots represent individual wells; data were analyzed by two-way ANOVA with Sidak’s correction for multiple comparisons. Unless otherwise stated, * p < 0.05, ** p < 0.01, ns. not significant.

The abundance of different lineages among splenocytes was analyzed by mass cytometry from 2- or 8-month-old mice in C9orf72^+/+ or C9orf72^−/− (Fig. 2B). We identified nine unique cell types based on their cell surface marker expression (Fig. 2, C and D and fig. S3, A to H). At 2 months of age, C9orf72^−/− mice displayed a significant accumulation of CD8⁺ DCs (p = 0.0040), CD8⁻ DCs (p = 0.0043), monocytes (p < 0.0001), neutrophils (p < 0.0001) and B cells (p = 0.0243) (Fig. 2D). By 8 months of age, C9orf72^−/− maintained these elevated cell populations and also displayed an increase in macrophages (p = 0.0214) and CD4⁺ T cells (p = 0.0005) (Fig. 2D). Thus, C9orf72 deficiency caused changes in multi-lineage hematopoiesis (11).

Next, we quantified the median staining intensity of each surface effector in the nine gated populations and performed a Bonferroni-corrected two-tailed t-test or Mann-Whitney test for non-normal distributions with significance threshold set to p < 0.0056 (0.05 divided by 9 to correct for multiple comparisons). This analysis revealed 44 significant differences between C9orf72^−/− and control populations at 2 months of age and 51 significant differences at 8 months of age (p < 0.0056) (Fig. 2E and fig. S3F). Increased surface expression of the high-affinity F_c γ Receptor (CD64) on C9orf72^−/− macrophages, CD8⁻ DCs, and monocytes, but not neutrophils (Fig. 2E), was consistent with development of autoimmune disease rather than acute bacterial infection in these animals (43). Surface expression of the chemokine receptor CXCR4 and hyaluronan receptor CD44, both involved in homing to sites of inflammation, were increased on multiple C9orf72^−/− populations (fig. S3F). Macrophages from C9orf72^−/− mice displayed a significant elevation in the surface expression of CD83 (p < 0.0056) (Fig. 2E), a molecule that is highly expressed in monocytes from patients with ALS with a rapid disease course (44). Additionally, C9orf72^−/− macrophages showed elevation of the immune checkpoint receptors leukocyte immunoglobulin-like receptor subfamily member 4 (LILRB4/CD85k) and Signal-regulatory protein α (SIRPα/CD172A), the immunosuppressive ectonucleotidase CD39, and the scavenger receptor CD48 (Fig. 2E). Together, these data suggest that the overall profile of C9orf72^−/− APCs is biased towards a pro-inflammatory state and that the balance of inhibitory molecules may govern the extent of disease in each mutant animal.

C9orf72 restricts co-stimulatory potential of myeloid cells in the periphery and central nervous system (CNS).

Naïve T cells require two signals to initiate a self-reactive inflammatory response. The first signal is provided by the interaction of the T cell receptor with a self-antigen that is displayed within the major histocompatibility complex (MHC) of APCs, whereas a second co-stimulatory signal licenses T cells to repeatedly proliferate (45). We found no difference in the abundance of MHC class I (H-2D^b/H-2L^d/H-2D^q/H-2L^q) or MHC class II (I-A/I-E) on the surface of C9orf72^−/− APCs relative to controls (Fig. 2E and fig. S3F). However, the co-stimulatory molecule CD80 was prominently enriched on the surface C9orf72^−/− macrophages and CD8⁺ DCs at 2 months of age and persisted on C9orf72^−/− macrophages at 8 months of age (Fig. 2E). The co-stimulatory molecule CD86 was consistently enriched on the surface of C9orf72^−/− CD8⁻ DCs, and the CD40 co-stimulatory molecule was enriched on C9orf72^−/− CD8⁺ DCs at 2 months of age (Fig. 2E).

Given that inflammatory phenotypes depend on the gut microbiota in this model (27), we interrogated co-stimulatory molecule expression on immune cells isolated from the colon, Peyer’s patch, mesenteric lymph node (mLN), and spleen of C9orf72^+/+ and C9orf72^−/− mice. Relative to C9orf72^+/+ controls, we observed a 54-fold enrichment of CD80^hi CD11b⁺ cells in the spleen (p = 0.0298) and 23-fold enrichment of CD80^hi CD11b⁺ cells in mLNs (p = 0.0042) of C9orf72^−/− mice, respectively, whereas no differences were observed in Peyer’s patch- or colon-derived myeloid cells (Fig. 2, F and G). To test the functional co-stimulatory potential of APCs, we treated tissue resident macrophages from C9orf72^−/− and C9orf72^+/+ mice with OVA_323-339 peptide in co-culture with OTII CD4⁺ T cells. Enrichment of CD80 on C9orf72^−/− macrophages from spleens and mLNs was associated with enhanced ability to promote antigen specific T cell proliferation (Fig. 2H).

We next asked whether co-stimulation was altered in the CNS of C9orf72 LOF mice. Evaluation of mass cytometry performed on forebrain and spinal cord of 8-month-old C9orf72^+/+, C9orf72^+/−, and C9orf72^−/− mice revealed significant accumulation of CD80^hi microglia in the forebrain and CD80^hi monocytes in the forebrain and spinal cord of C9orf72^−/− mice relative to C9orf72^+/+ controls (p < 0.05) (Fig. 3, A and B). No enrichment of CD80 was observed in C9orf72^+/− CNS or periphery (Fig. 3, A and B and fig. S3G). MHC-II^hi myeloid cells were elevated in C9orf72^−/− forebrain and spinal cord, however, CD86, CD40, and MHC I did not differ between genotypes (fig. S4A). Thus, C9orf72 restrains myeloid co-stimulation in the brain and secondary lymphoid organs, where CD80 may enhance antigen-specific helper T cell responses.

Fig. 3. CD80 is enriched in microglia from C9orf72^−/− mice and patients with C9ORF72 ALS.

(A and B) Mass cytometry of forebrain and spinal cord of male and female littermates at 8 months of age (C9orf72^+/+, n=7; C9orf72^+/−, n=7; C9orf72^−/−, n=8). One C9orf72^+/+ forebrain sample was excluded due to failed tissue isolation. (A) Representative CD80 and CD86 expression in microglia (left, CD45^mid CD11b⁺ CD39⁺ CX3CR1⁺) and monocytes (right, CD45^hi CD11b⁺ Ly6C⁺). Numbers on plots represent percent of the gated population. (B) Abundance of each gated population. Each dot represents one animal; data were analyzed by one-way ANOVA with Dunnett’s correction for multiple comparisons. (C) Representative human spinal cords stained for Iba1 (green), CD80 (red), and DAPI (blue); scale 10μm. (D) CD80 intensity per Iba1⁺ human spinal cord microglia with n > 440 microglia imaged per case (left axis, violin plot). Overlaid dot represents the average microglia CD80 intensity per case (right axis). Data were analyzed by unpaired Student’s t-test of average microglia CD80 intensity between C9ORF72 ALS cases (n=4) and non-ALS controls (n=3). For all panels: * p < 0.05, ** p < 0.01, ns. not significant.

Human C9ORF72 ALS microglia accumulate CD80.

To determine whether high CD80 expression was a conserved feature in ALS, we stained human spinal cord from C9ORF72 ALS cases (n=4) and non-ALS controls (n=3) for CD80 and imaged more than 440 Iba1⁺ microglia per case (Fig. 3C). We observed significantly greater CD80 staining on C9ORF72 ALS spinal cord microglia relative to non-ALS microglia (p = 0.0199) (Fig. 3, C and D and fig. S4B and table S3). Enrichment of CD80 on C9orf72^−/− microglia, an outcome not observed in C9orf72^+/− mice, modeled the phenotype in human ALS spinal cord.

C9ORF72 restrains CD80 trafficking in response to inflammatory stimuli.

Protein abundance on the cell surface depends, in part, on the dynamic balance between externalization and internalization of vesicles. To glean additional insight into regulation of co-stimulation by C9orf72, we first tested whether C9orf72 LOF affected receptor-mediated endocytosis of CD80 or the lipopolysaccharide (LPS) sensor toll-like receptor 4 (TLR4). Splenic macrophages from C9orf72^−/− mice showed no deficit in CD80 or TLR4 internalization relative to C9orf72^+/+ cells (fig. S5, A and B). Next, we used CRISPR/Cas9 to edit C9orf72 in Raw 264.7 mouse macrophages (C9orf72^Δ) and compared these cells with unedited isogenic controls (C9orf72^WT). We found that CD80 accumulated to a greater extent on the surface of C9orf72^Δ cells than C9orf72^WT cells in response to the TLR2 agonist Pam₃csk₄ or the TLR7/8 agonist R848, but CD80 was not elevated to a greater extent in response to LPS or the TLR3 agonist poly(I:C) (Fig. 4A and fig. S5, C and D). MHCII surface accumulation was not potentiated by C9orf72 deficiency (Fig. 4A and fig. S5, C and D). The differential responsiveness to TLR-dependent stimuli may explain why C9orf72^−/− macrophages accumulate CD80, but not MHCII, on their cell surface (Fig. 2E).

Fig. 4. C9orf72 governs CD80 trafficking in response to TLR agonist and cytokine exposure.

CRISPR/Cas9-edited Raw 264.7 murine macrophage cells with intact (C9orf72^WT) or LOF mutated (C9orf72^Δ) C9orf72 locus were used for in vitro studies in all panels. Each dot represents one well. (A) Surface antibody staining and flow cytometry for CD80 or MHCII after 18hr exposure to vehicle or TLR2 agonist Pam₃csk₄. (B) Whole cell lysate western blot and densitometry quantification analysis 18hrs after exposure to vehicle or Pam₃csk₄. (C) Immunofluorescent staining for CD80 (red) after 18hr exposure to vehicle or Pam₃csk₄. (D) Experimental design with PIKFyve lipid kinase inhibitor Apilimod to disrupt vesicle trafficking. (E) Surface or (F) intracellular CD80 staining intensity by flow cytometry. (G) Whole cell lysate western blot and quantification after 18hr exposure to vehicle or recombinant murine IFN-γ with or without recombinant murine IL-17A. (H) Surface antibody staining for CD80 and flow cytometry after 18hr vehicle or cytokine exposure. (I) C9orf72^Δ cells transfected with GFP or human C9ORF72 expression plasmid 2 days before 18hr stimulation, surface staining for CD80 and analysis by flow cytometry. Each dot represents one replicate; all experiments were repeated at least twice; data were analyzed by two-way ANOVA with Sidak’s correction for multiple comparisons; * p < 0.05. ** p < 0.01. ns. not significant.

Following exposure to Pam₃csk₄, CD80 was enriched in whole cell lysate of C9orf72^Δ cells relative to C9orf72^WT controls (Fig. 4B) and immunostaining revealed that CD80 accumulated on the perimeter of stimulated C9orf72^Δ cells (Fig. 4C). Given that C9ORF72 can act as a GAP for small GTPases involved in actin cytoskeleton and vesicle movement (6–8), we hypothesized the altered abundance of CD80 in C9orf72^Δ cells was due to aberrant vesicle trafficking. To test this, we added Apilimod, an inhibitor of the lipid kinase enzyme PIKfyve that regulates endomembrane homeostasis (9), to cells 3 hours after Pam₃csk₄, when most changes in TLR signaling and CD80 expression had already occurred. We then measured CD80 by flow cytometry (Fig. 4D). Apilimod treatment caused CD80 to accumulate inside C9orf72^Δ cells rather than on their surface (Fig. 4, E and F), suggesting that CD80 traffics through PIKfyve-dependent vesicles in myeloid cells.

Next, we considered whether inflammatory cytokines enriched in C9orf72^−/− plasma could elicit aberrant co-stimulation. Recombinant interferon (IFN)-γ exposure caused C9orf72^Δ mouse macrophages, but not C9orf72^WT cells, to accumulate CD80 in total lysate and on the surface (Fig. 4, G and H). Whereas recombinant IL-17A (rIL-17A) alone did not affect expression of CD80, co-administration of rIL-17A with IFN-γ further enhanced CD80 in a C9orf72-dependent manner (Fig. 4, G and H). Moreover, introduction of a human C9ORF72-expressing plasmid prior to cytokine stimulation reduced CD80 induction on C9orf72^Δ cells (Fig. 4I). Together these findings suggest that C9ORF72 restrains CD80 trafficking to the cell surface in response to microbial-derived agonists of TLR2, TLR7, and TLR8 as well as IL-17A and IFN-γ, cytokines produced by pathogenic Th17 cells.

Induction of lysosomal exocytosis by the lysotrophic agent chloroquine was impaired in C9orf72^Δ cells relative to C9orf72^WT cells, as illustrated by the reduced abundance of Lamp1 on the cell surface (fig. S5E). The muted response of C9orf72^Δ cells to chloroquine correlated with elevation of lysosomal pH (fig. S5F) and is consistent with reduced protonation of chloroquine that leads to its entrapment in the organelle (46). After TLR2 stimulation, C9orf72^Δ cells were more likely to adopt a CD80^hi Lamp1^lo phenotype, rather than a CD80^lo Lamp1^hi phenotype (fig. S5, G and H). This raised the possibility that C9orf72 governs the transition to unique myeloid cell states. In support, we found that CD80^hi Lamp1^lo and CD80^lo Lamp1^hi splenic macrophages were elevated in C9orf72^−/− mice and C9orf72^2Lox/2Lox;LysM-Cre+ mice compared with age matched controls (fig. S5, I and J). Future efforts will delineate how these divergent macrophage populations contribute to organismal health and disease.

Microglia and endothelial cells participate in C9orf72 LOF-dependent neuroinflammation.

C9orf72^−/− mice develop neuroinflammatory features that resemble C9ORF72-ALS/FTD (10, 11, 27, 28). To explore the temporal and anatomic basis of these phenotypes, we performed single-cell RNA sequencing of the dissociated cortex and lumbar spinal cord from C9orf72^+/+ and C9orf72^−/− littermates at 2, 4, and 8 months of age (Fig. 5A). Quality control yielded 50,578 sequenced cells for downstream analysis. We used Seurat (47) to separate cells into clusters with t-stochastic neighbor embedding (tSNE) dimensionality reduction and identified microglia (complement C1q subunit A (C1qa)), endothelial cells (claudin-5 (Cldn5), vascular endothelial growth factor receptor 1 (Flt1)), astrocytes (connexin 43 (Gja1)), pericytes (proteolipid protein 1 (Plp1)), oligodendrocytes (cyclic nucleotide phosphodiesterase (Cnp)), and neurons (doublecourtin (Dcx), synaptosomal-associated protein 25 (Snap25), stathmin 2 (Stmn2)) from each library (fig. S6, A to D) and found the proportion of each cell type consistent across time points, tissues, and genotypes (fig. S6E).

Fig. 5. C9orf72^−/− microglia and endothelial cells exhibit a coordinated inflammatory gene profile.

(A) Single cell RNA sequencing workflow from spinal cord and brain cortex at 2, 4 and 8 months of age in female C9orf72^+/+ (n=3) and C9orf72^−/− (n=3) littermates. (B) Total number of genes differentially expressed (adjusted p < 0.05) between C9orf72^+/+ and C9orf72^−/− populations. Up-regulated genes are red, down-regulated genes are blue. Abbreviations include endothelia (Endo), microglia (Micro), astrocyte (Astro), pericyte (Peri), oligodendrocyte (Oligo), spinal cord (Crd), brain cortex (Ctx). (C to E) Module score of genes upregulated in endothelial cells (C), microglia (D), or astrocytes (E) across C9orf72^−/− (red) and C9orf72^+/+ (gray) populations. (F and G) gProfiler pathway enrichment based on genes increased in C9orf72^−/− (F) microglia or (G) endothelial cells. (H and I) gProfiler pathway enrichment based on genes decreased in C9orf72^−/− (H) microglia or (I) endothelial cells. Gene Ontology (GO); Biological Pathways (BP); Cell Compartment (CC); Kyoto Encyclopedia of Genes and Genomes (KEGG). (J) Genes differentially expressed between C9orf72^+/+ and C9orf72^−/− microglia in spinal cord and cortex by age. (K) Genes differentially expressed between C9orf72^+/+ and C9orf72^−/− endothelial cells in spinal cord and cortex by age.

The greatest number of significant differentially expressed genes (DEGs, p< 0.05) occurred between C9orf72^+/+ and C9orf72^−/− endothelial cells (236 spinal cord DEGs; 264 cortex DEGs), followed by microglia (145 spinal cord DEGs; 202 cortex DEGs), astrocytes (90 spinal cord DEGs; 110 cortex DEGs), pericytes (25 spinal cord DEGs; 45 cortex DEGs) and oligodendrocytes (39 spinal cord DEGs; 0 cortex DEGs) (Fig. 5B and data file S1). Relatively few neurons were sequenced after myelin removal, so these cells were excluded from the DEG analysis. To determine whether C9orf72 regulates a shared set of genes in distinct cell types of the brain, we generated a z-score of all DEGs elevated in each cell population and compared the expression of this module across cellular populations in the dataset. Genes that were upregulated in C9orf72^−/− endothelial cells showed elevated expression in microglia and pericytes (Fig. 5C). Reciprocally, genes upregulated in C9orf72^−/− microglia were modestly upregulated in endothelial cells (Fig. 5D). In contrast, genes upregulated in C9orf72^−/− astrocytes were expressed to a similar extent across the other cell populations profiled (Fig. 5E).

Gene ontology (GO) analysis of DEGs in each population demonstrated that C9orf72^−/− microglia and endothelial cells displayed concordant changes in pathways of immune system processes, such as antigen processing and presentation, defense response to other organisms, and response to viruses (Fig. 5, F and G and fig. S6, F and G). Pathways elevated in C9orf72^−/− astrocytes related to peptide metabolism, glycolysis, and myelin sheath (fig. S6H). We note that C9orf72^−/− microglia, endothelial cells, and astrocytes all exhibited downregulation of genes involved in oxidative phosphorylation (mt-Atp6, mt-Co1, mt-Co3) (Fig. 5, H to K and fig. S6, I and J), consistent with a previously reported role for C9ORF72 in regulation of the electron transport chain (48). At 2 months of age, C9orf72^−/− microglia and endothelial cells displayed significant elevation of MHCI genes (H2-D1, H2-T23, H2-K1), whereas at 8 months of age we found a shared elevation of interferon-inducible genes (Ifi27l2a, Ifitm3, Slfn2) (p < 0.05) (Fig. 5, J and K). At 4 months of age, C9orf72^−/− microglia expressed chemokines (Ccl2, Ccl3, Ccl4) and cytokines (Il-1b, Il-1a, Tnf) that attract and activate peripheral immune cells (Fig. 5J), whereas C9orf72^−/− endothelial cells expressed genes associated with circulatory system development (Tfrc, Sox18, Ankrd37) (Fig. 5K).

Lall and colleagues (49) performed single cell transcriptomics of cortex from C9orf72^−/− mice and identified an interferon response microglia (IRM) and activated response microglia (ARM) gene signature at 17 months of age in C9orf72^−/− animals. We found 13 IRM genes and 33 ARM genes to be differentially expressed in C9orf72^−/− microglia from our study (fig. S7, A and B). Microglial pathways shared between the studies included viral process, defense response to other organisms, and response to type II interferon (fig. S7C). Elevation of the IRM gene module in our study was greatest in C9orf72^−/− microglia at 8 months of age (fig. S7, D and E). Comparison of endothelial cells between the studies revealed 53 concordant upregulated DEGs and 12 shared downregulated DEGs (fig. S7, F and G) and shared pathways included antigen processing and presentation, defense response to other organisms, and response to type II interferon (fig. S7H).

To determine whether the changes in gene expression we observed in the CNS were conserved in peripheral immune cells, we performed single cell RNA sequencing of C9orf72^−/− and C9orf72^+/+ splenocytes and identified macrophages (Lyz2), conventional DCs (Itgax), plasmacytoid DCs (Siglec-H), neutrophils (Ly6G), T cells (Cd3e, Cd4, Cd8a), B cells (Cd19), plasma cells (Sdc1) and red blood cells (Hba-a1) (fig. S8, A and B). C9orf72^−/− myeloid and lymphoid cell populations exhibited concordant changes in genes associated with ALS, regulation of organelle organization, and antigen processing and presentation that mirrored alterations in microglia (fig. S8, C to F). We validated that Apolipoprotein E (Apoe) and Cathepsin B (Ctsb) are elevated in C9orf72^−/− macrophages (fig. S8, G and H and table S4). These data suggest that C9orf72 regulates overlapping gene programs in the CNS and the periphery, potentially in response to inflammatory cues.

IL-17A promotes C9orf72^−/− dependent systemic inflammation and CD80 co-stimulation.

We hypothesized that IL-17A is an important mediator of inflammation in C9orf72^−/− mice because transplantation of C9orf72^−/− bone marrow into C9orf72^+/+ mice caused IL-17A accumulation and fatal autoinflammatory disease (11), whereas treatment of C9orf72^−/− mice with antibiotics or fecal microbial transplantation reduced IL-17A abundance and ameliorated autoimmune and inflammatory phenotypes (27). To test this hypothesis, we generated and intercrossed C9orf72^+/− Il-17a^+/− mice from different breeding pairs to make littermates for study (fig. S9A). We observed no overt inflammation in C9orf72^+/+ mice independent of the copies of Il-17a (fig. S9, B to F), so these animals were pooled (C9orf72^+/+;Il-17a^all) for comparisons. Relative to C9orf72^+/+;Il-17a^all controls, C9orf72^−/−;Il-17a^+/+ mice developed features of systemic inflammation that included neutrophilia (fig S9B), reduced platelet count (fig. S9C), splenomegaly (Fig. 6A, fig. S9D), and mLN hyperplasia (Fig. 6B, fig. S9E). In contrast, homozygous deletion of Il-17a in C9orf72^−/− mice alleviated lymphocytosis and neutrophilia in spleens and mLNs (Fig. 6, A and B). Th17 cells and T_reg cells were reduced in C9orf72^−/−;Il-17a^−/− mLNs but not spleens relative to C9orf72^−/−;Il-17a^+/+ mice (fig. S9, G to J). Moreover, Il-17a deficiency prevented the accumulation of C9orf72^−/− CD80^hi myeloid cells in mLN and spinal cord (Fig. 6, C to E and fig. S9K) and reduced IgM but not IgG autoantibodies in C9orf72^−/− plasma (fig S10A and B). Thus, these data suggest that in the absence of C9ORF72, IL-17A augments CD80 co-stimulation in the gut and CNS.

Fig. 6. Il-17a deficiency reduces CD80 expression in C9orf72^−/− gut and spinal cord.

(A to C) Cellular abundance of each gated population at 6 months of age. (A) Ly6G⁺ Neutrophil, CD4⁺ T cell, and CD19⁺ B cell abundance per spleen. (B) Ly6G⁺ Neutrophil, CD4⁺ T cell, and CD19⁺ B cell abundance per mLN. (D) Spleen and mLN CD11b⁺ CD80^hi cell abundance. (E) Representative surface staining of CD80 on CD11b⁺ spinal cord cells. (F) Spinal cord CD45^mid CD11b⁺ CD80^hi microglia and CD45^hi CD11b⁺ CD80^hi monocyte/neutrophil abundance. C9orf72^+/+ Il-17a^+/+ (n=4 per group); C9orf72^+/+Il-17a^+/− (n=4 per group); C9orf72^+/+ Il-17a^−/− (n=4 per group); C9orf72^−/− Il-17a^+/+ (n=5 per group); C9orf72 ^−/− Il-17a ^+/− (n=7 per group); C9orf72 ^−/− Il-17a ^−/− (n=3 per group) with sexes combined. One C9orf72^−/− Il-17a^+/− spinal cord sample was excluded due to failed isolation of cells within it; each dot represents one mouse; horizontal bars represents mean; data were analyzed by one-way ANOVA with Dunnett’s correction multiple comparisons within spleen Neutrophil, CD4⁺, B cell, mLN Neutrophil, B cell, CD80^hi CD11b⁺ and spinal cord CD45^hi CD11b⁺ CD80^hi monocyte/neutrophil or Kruskal-Wallis test with Dunn’s multiple comparisons within spleen CD80^hi CD11b⁺, mLN CD4⁺ and spinal cord CD45^mid CD11b⁺ CD80^hi microglia ; * p < 0.05, ** p < 0.01, ns. not significant.

IL-17A neutralization improves motor function and neuronal health in C9orf72^−/− mice.

Lastly, we investigated whether IL-17A neutralization could reverse established neuroinflammation in C9orf72^−/− mice. We aged an additional cohort of C9orf72^+/+ (n=38; 22 female:16 male) and C9orf72^−/− mice (n=36; 11 female:25 male) and distributed mutants into treatment groups based on rotarod performance and blood measurements (fig. S11, A to D). Then we began treatment with vehicle (C9orf72^+/+ vehicle, n=22 female, n=16 male), isotype control antibody (C9orf72^−/− isotype, n=6 female, n=12 male) or anti-IL-17A neutralizing antibody (C9orf72^−/− anti-IL-17A, n=5 female, n=13 male), followed by analysis of peripheral organ and brain inflammation after 6 weeks (Fig. 7A). We found that improvement of motor function over the study duration was significantly greater in C9orf72^−/− anti-IL-17A females (Fig. 7B; p = 0.0022, one-way ANOVA with Tukey’s correction for multiple comparisons) and males (Fig. 7C; p = 0.0343, one-way ANOVA with Tukey’s correction for multiple comparisons) relative to C9orf72^+/+ vehicle controls. In contrast, no difference in motor improvement was observed between C9orf72^+/+ vehicle and C9orf72 ^−/− isotype mice (Fig. 7, B and C). Circulating neutrophils were reduced in female, but not male, C9orf72^−/− anti-IL-17A mice relative to C9orf72^−/− isotype controls (Fig. 7, D and E) although other measures of peripheral inflammation did not differ between C9orf72^−/− anti-IL-17A and C9orf72^−/− isotype groups (fig. S11, E to J). Moreover, no deaths occurred in the C9orf72 ^+/+ vehicle group; however, two C9orf72^−/− isotype males and two C9orf72^−/− anti-IL-17A females died over the treatment course (fig. S11, K to M). Finally, spinal cord cells from male and female C9orf72^−/− isotype mice displayed elevated expression of Apoe, Histocompatibility 2-Q7 region (H2-Q7), Interferon induced transmembrane protein 3 (Ifitm3), and Interleukin 1 beta (Il1b); in contrast, this neuroinflammatory gene signature was not different between C9orf72^−/− anti-IL-17A cells relative to C9orf72^+/+ vehicle controls and Apoe was lower in C9orf72^−/− anti-IL-17A compared with C9orf72^−/− isotype (Fig.7F).

Fig. 7. Therapeutic neutralization of IL-17A improves motor performance in C9orf72^−/− mice.

(A) Study design with female (open symbol) and male (closed symbol) C9orf72^+/+ (n=22 female and n=16 male) and C9orf72^−/− (n=11 female and n=25 male) mice. Treatment groups included C9orf72^+/+ mice treated with vehicle (n=38), C9orf72^−/− mice treated with isotype control antibody (n=6 female and n=12 male; n=2 males not assessed due to premature mortality) and C9orf72^−/− mice treated with anti-IL-17A antibody (n=5 female and n=13 male; n=1 female not assessed due to premature mortality). (B and C) Rotarod performance of (B) females and (C) males over study duration. Data presented as mean ± SEM. For bar plots, each dot represents average fold change in performance of individual mice over 6 weeks relative to pre-treatment; data were analyzed by one-way ANOVA with Tukey’s correction for multiple comparisons. (D and E) Neutrophil count in peripheral blood of (D) females and (E) males over study duration. For bar plots, each dot represents individual mice at week 6 time point; data were analyzed by (D) one-way ANOVA with Tukey’s correction for multiple comparisons and (E) Kruskal-Wallis test with Dunn’s correction for multiple comparisons. (F) Normalized gene expression of single cell dissociated myelin-depleted spinal cord cells evaluated in the following animals: C9orf72^+/+ mice treated with vehicle (n=6 female and n=9 male), C9orf72^−/− mice treated with isotype control antibody (n=3 female and n=6 male) and C9orf72^−/− mice treated with anti-IL-17A antibody (n=3 female and n=7 male). Each dot represents one mouse; horizontal bar represents mean; data were analyzed by two-way ANOVA with Tukey’s correction for multiple comparisons; for all panels: * p < 0.05, ** p < 0.01, ns. not significant.

DISCUSSION

This work sought to address two interrelated questions: To what extent are the immune phenotypes observed in C9orf72-deficient mice attributable to the function of this gene in hematopoietic cells and through what mechanism does C9ORF72 promote organismal health? Here we report that C9orf72 acts within the myeloid lineage, particularly in macrophages and monocytes, to prevent the initiation of autoimmune disease. Our findings support a model in which C9ORF72 normally restrains CD80 surface expression within myeloid cells following exposure to inflammatory stimuli. However, when functional C9orf72 expression declines, CD80 is aberrantly trafficked to the cell surface, where it provides an antigen-independent co-stimulatory signal that promotes activation of T cells, including those with reactivity to self-antigens. Over time, this chronic co-stimulation leads to broad loss of tolerance and premature mortality. In addition, our findings revealed a complementary role for C9orf72 to limit IL-17A-dependent inflammation. Specifically, IL-17A acts in a feed-forward manner to enhance myeloid CD80 co-stimulation in the spinal cord and draining lymph nodes of the gut, where it enables T helper cell differentiation.

The IL-17A inflammatory axis may be a convergent pathologic mechanism across diverse forms of ALS/FTD. As a biomarker, IL-17A is enriched in the serum and spinal fluid of patients with ALS/FTD, including those with a C9ORF72 mutation (23–26). In addition, Th17 cells are present at a higher frequency in the blood of patients with ALS at disease onset and correlate with a faster rate of progression (21, 50, 51). IL-17A may act directly to induce neural degeneration (52, 53), or indirectly by disruption of brain vasculature (54) or the promotion of neurotoxic cells such as neutrophils (55). Indeed, neutrophils are enriched in the blood and cerebral spinal fluid of patients with sporadic ALS and neutrophilia negatively correlates with rate of survival (56, 57). Our demonstration that C9ORF72 ALS microglia increase expression of CD80, that IL-17A and IFN-γ induce CD80 expression, and that depletion of Il-17a attenuates CD80 in the spinal cord of C9orf72-deficient mice further elaborate how IL-17A and aberrant co-stimulation (58) might exacerbate motor neuron disease. Encouragingly for patients, therapies that target IL-17A are approved by the United States Food and Drug Administration (59) and could be quickly repurposed to treat ALS/FTD.

In addition to governing IL-17A production, we found that C9orf72-LOF associated with altered B cell immunoglobulin class switching. The transition from IgM to IgG is important for anti-viral and anti-cancer immunity, in part due to an increase in the affinity of the antibodies for antigen as well as enhanced F_cγ receptor IIIa-mediated cytotoxicity (60). The regulation of class switch recombination is complex and involves B cell-intrinsic signals and extrinsic signals such as cytokines provided by APC-dependent helper T cells. Similarly, the decision of naïve T cells to differentiate into T helper type 1, Th17 or T_reg cells or to trans-differentiate during chronic immune responses (61) is dependent on the strength of T cell receptor signaling as well as the local milieu of cytokines and ligands present (62, 63). Future efforts will focus on molecular regulation of B cell class switch recombination and T cell fate determination by C9ORF72.

Bulk RNA sequencing of brain samples from patients with C9ORF72 ALS has demonstrated an enrichment of genes related to antigen processing and presentation and anti-viral immunity (28, 64). Our finding that microglia from C9orf72-deficient mice express a similar signature is consistent with previous studies that have profiled the nervous system of these mutant mice using bulk and single cell RNA sequencing (10, 49). By extending our single cell analysis to other glial populations, we discovered concordantly enriched gene programs in C9orf72-deficient endothelial cells. This convergence may have occurred, in part, as a response to ongoing systemic inflammation and disruption of the blood spinal cord barrier (27) as evidenced by increased titers of autoantibodies against Aquaporin 4, as well as evidence of increased activation of the coagulation and complement cascades. Since endothelial cells share a common developmental origin with Vav1+ hematopoietic stem cells (31), induce Mx1 in response to interferon (32), and express STING (28, 65), our findings raise the possibility that C9orf72 may act within the brain vasculature to promote tolerance to neural antigens. An important future experiment will be to determine whether rescuing C9ORF72 expression in the neurovascular unit can improve blood spinal cord barrier integrity that becomes disrupted in patients with ALS (66, 67).

Our study is not without limitations. Although we measured CD80 expression in spinal cord from four patients with C9ORF72 ALS and three non-ALS controls, a larger survey of sporadic and familial ALS and FTD patients will help to determine whether microglia CD80 is a general feature of the disease or a specific consequence of the C9ORF72 mutation. To elucidate the relationship between CD80 co-stimulation, motor neuron loss, and rate of motor decline, evaluation of CD80 across affected and unaffected brain regions of patients annotated for fast or slow disease progression should be performed. Whereas IL-17A neutralization improved motor function and suppressed neuroinflammation in aged C9orf72^−/− mice, this therapy was not sufficient to reduce spleen weight or increase platelet counts. Prophylactic neutralization of IL-17A at a younger age may be required to prevent or reverse systemic inflammation and autoimmunity in this model. Moreover, by following the natural history of hundreds of mutant mice, we found that C9orf72^−/− males tended to die at younger ages than C9orf72^−/− females. A male predominance is also observed in ALS but this sex-bias is incompletely understood (68). Our observation that IL-17A neutralization suppressed circulating neutrophils in female, but not male, C9orf72^−/− mice may support a role for estrogen or X-linked genes in the regulation of IL-17-mediated granulopoiesis. In conclusion, our study has identified an important role for C9orf72 in myeloid and lymphoid cells to oppose autoimmunity and implicates IL-17A as a key effector of neuroinflammation in C9orf72-deficient mice. The role of C9ORF72 haploinsufficiency in ALS/FTD is still unclear and is likely to act in conjunction with environmental exposures (16, 27, 69), other ALS and FTD risk alleles (70), and GOF effects of the C9ORF72 mutation. Ultimately, a cure for ALS and FTD may involve a combination of approaches such as correction of the causal gene mutation, augmentation of protein and metabolic homeostasis, and reorientation of the immune system to promote neural protection and tissue regeneration. IL-17A neutralization may be one such complementary approach for ALS and FTD.

MATERIALS AND METHODS

Study design

The goal of this study was to identify cell types and pathways that contribute to autoimmunity and inflammation in C9orf72 LOF mice. The experimental design involved age-matched evaluation of C9orf72 mutant and control mice paired with mass cytometry, flow cytometry, single cell sequencing, antibody-based therapeutic neutralization, and cellular and biofluidic analyses. Sample sizes were determined based on power calculations. Pre-defined euthanasia criteria and age for euthanasia was determined based on temporal characterization of systemic and neuroinflammation in this model. Neutrophil counts, platelet counts, and rotarod performance were used to assign mice into isotype antibody or anti-IL-17A antibody treatment groups so these disease measures did not differ between C9or72^−/− mice prior to treatment. All subjects were coded so operators remained blinded to genotype during behavioral testing, tissue harvest, and outcome assessments. Numbers of mice and well replicates are indicated in figure legends.

Animals

All experimental protocols and procedures were approved by the Institutional Animal Care and Use Committees of Harvard University and Case Western Reserve University and were in compliance with all relevant ethical regulations. Two independent cohorts of C9orf72 LOF mice, initially developed and characterized at Harvard University Biological Research Infrastructure Building (HU BRI) (11, 27), were used this study. To establish a colony at Case Western Reserve University Wolstein Research Building (CWRU WRB), C9orf72^+/− mice were backcrossed onto C57BL6/J for six generations at HU and were rederived by aseptic embryo transfer to generate founders for propagation. The fecal bacteria of mice at CWRU WRB more closely resembled mice reared in HU rather than C57BL6/J mice from Jackson Laboratory (10, 27) and C9orf72^−/− developed inflammation and autoimmunity at CWRU WRB (fig. S12, A to E). Mice were housed with nestlet bedding on a 12-h light/dark cycle and provided ad libitum water and food (Prolab Isopro RMH 3000 (HU BRI) or Envigo Mouse/Rat Diet (CWRU WRB)). Embryo re-derivation was performed by collecting embryos from super-ovulated C9orf72^+/− females, washing embryos, then surgical transfer using aseptic technique into the reproductive tract of pseudo-pregnant recipient females. The generation of C9orf72 LOF animals with neomycin resistance cassette removed (neo deleted)was previously described (11). C9orf72 conditional mice were obtained by material transfer agreement from Dr. Jeroen Pasterkamp (30). Il-17a knockout mice were a kind gift from Dr. Nicole Ward (71). Cre lines used in this study were obtained from Jackson labs and included Vav1-cre (31), Mx1-cre (32)), Lyz2/LysM-cre (33), CD2-cre (34), CD19-cre (35), CD4-cre (36) and FoxP3-cre (37). Tail DNA was lysed in Viagen digest buffer (102-T) supplemented with proteinase K (Roche) for 12 hours at 55°C. Genotyping PCR was performed using Amplitaq gold (Thermo) as outlined for each strain (fig. S1D).

Human tissue samples

Post-mortem frozen spinal cord sections from control individuals and C9ORF72 carriers were obtained as previously described (78). This Health Insurance Portability and Accountability Act-compliant prospective study was approved by the Cleveland Clinic institutional review board and written informed consent was obtained from all participants. Eligibility required a diagnosis of ALS according to the revised El Escorial criteria.

Single cell sequencing

For dissociation of brain and spinal cord tissue for single cell sequencing, 2 mL glass pipette tips were used to homogenize the tissue and dissociated with papain (Worthington) supplemented with DNAse in Ca²⁺-free serum-free Earle’s Balanced Salt Solution pre-heated to 37°C. Tissue was dissociated at 37°C for 25-min, with homogenization after 10-mins using 1 mL plastic tip. After 25-min, the tissue was homogenized again with 1 mL plastic tip, filtered (70 μm), Dulbecco’s Modified Eagle Medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (FBS) added, then centrifuged at 300g for 10-min. Cells were subjected to myelin removal beads (Miltenyi), counted, diluted in Ca²⁺-free HBSS with 0.5% Bovine Serum Albumin, and loaded into a Chromium Single Cell 3’ Chip (10X Genomics) and processed following the manufacturer’s instructions. Sequencing libraries were prepared with the Chromium Single Cell 3’ library and Gel Bead kit v2 (10X Genomics). Libraries were sequenced on a Nova-seq 6000. Reads were aligned to the mouse genome assembly GRCm38. Count matrices were generated for each sample, and a combined UMI matrix was generated using merge. This matrix was first filtered for number of genes (nFeature) and percentage of mitochondrial and ribosomal genes (400< nFeature_RNA < 4000, percent.mito < 25, percent.RPS < 15, percent.RPL < 15) and then normalized and scaled manually using default Seurat algorithm (v3.0.1) (47). After quality filtering, filtered barcodes were used to compute Shared nearest-neighbor graphs and t-SNE projections using the first 8 statistically significant Principal Components, then cell identity was assigned on broad cell type specific markers as previously described (72). Analysis of cellular subtypes were conducted by subsetting each cell type as separate group. Isolated barcodes were re-normalised and scaled and relevant PCs were used for re-clustering as a separate analysis. This newly scaled matrix was used for Differential Gene Expression analysis. gProfiler used to quantify KEGG and Gene ontology pathway enrichment of differentially expressed genes for each identified cluster. For spleens, Cell Ranger 2.2.0 used to process raw sequencing data to convert Illumina basecall files to fastq, aligned sequencing reads to mm10 transcriptome using STAR aligner (73) and quantified the expression of transcripts in each cell. Analysis of processed scRNAseq data in R using Seurat (74, 75) and tidyverse packages (76). K means clustering applied to assigned clusters using SPRING (77). Log fold enrichment of each gene within identified clusters mapped using Morpheus (https://software.broadinstitute.org/morpheus).

Macrophage and OTII T cell cocultures

Macrophages were isolated from single cell dissociated mLNs, spleens and colons using F4/80 positive selection beads (Miltenyi) as per the manufacturer’s instructions. Macrophages were counted and plated in ventilated tubes in an incubator for 18 hours in T cell co-culture media containing RPMI 1640, 10% heat inactivated FBS (high grade Hyclone SH30084.03HI), 1% Sodium Pyruvate, 50 μM β-mercaptoethanol, 10 mM HEPES, 5 mL penicillin-streptomycin and Glutamax in the presence of 50 μg/mL OVA_323-339 peptide (Invivogen vac-isq). The following day, macrophages were centrifuged, washed in fresh media twice, and 5,000 cells per well were plated into 96 well round bottom tubes. CD4⁺ CD25⁻ OTII splenic T cells were bead enriched (Miltenyi), counted, and loaded with CFSE at 5 μg/mL for 20 mins in phosphate buffered saline (PBS); cells were then washed in complete medium, counted, diluted in T cell co-culture media supplemented with recombinant human IL-2 (Peprotech) (2.5 ng/mL), and plated at 40,000 cells per well onto the OVA-loaded macrophages. After 4 days in culture, cells were analyzed for CSFE dilution on a BD Fortessa.

IL-17A neutralization and motor behavior

Naïve animals were trained on the rotarod at constant speed of 4 RPM for 300 seconds at least one day before competitive assessment. For performance trials, the rotarod accelerated from 4 to 40 RPM over 300 seconds using Ugo Basile mouse RotaRod NG. Each trial day consisted of three tests per mouse, with each test separated by at least 20 minutes. Operator was blinded to animal genotype during trials. During the study period vehicle, anti-mouse IL-17A or IgG1 isotype control (InVivoMAb) was administered at a dose of 30 mg/kg intraperitoneally twice per week.

Statistical Analysis

A majority of the data was analyzed using Prism 10 (GraphPad Inc.). Unpaired t-test was used to test for significance between two groups. For distributions that did not meet the assumption of normality using Kolmogorov-Smirnov test, we applied the Mann-Whitney U test. For comparisons across multiple conditions one- or two-way ANOVA was performed with Sidak’s, Tukey’s, or Dunnett’s correction for multiple comparisons or Kruskal-Wallis test with Dunn’s correction for multiple comparisons when data did not meet the assumption of normality using Kolmogorov-Smirnov test. Figure legends indicate each statistical test and n values that refer to the number of biological or well to well replicates. All cell culture experiments were repeated at least twice. Differences between groups were deemed significant at two-tailed p < 0.05. For comparison of surface antigen expression across spleen populations (CyTOF), a Bonferroni corrected T-test or Mann-Whitney test for non-normal distributions was performed to compare 2 groups: C9orf72^+/+ and C9orf72^−/−. The n=9 cell types of the spleen identified were treated as subgroups, so the Bonferroni correction was calculated as alpha divided by n or 0.05/9 = 0.0056. Single cell sequencing data were analyzed in R using the Seurat pipeline and genes expressed in >25% of cells with Log₂ Fold change >0.25 adjusted p-values <0.05 were considered significant. Survival data were assessed by Gehan-Breslow-Wilcoxon test.

Data and materials availability:

All data associated with this study are in the paper or the supplementary materials. RNA sequencing datasets can be assessed through GEO accession #s GSE252888. C9orf72^2Lox mice were obtained from Dr. Jeroen Pasterkamp through material transfer agreement.