C9orf72 ablation causes immune dysregulation characterized by leukocyte expansion, autoantibody production, and glomerulonephropathy in mice

The expansion of a hexanucleotide (GGGGCC) repeat in C9ORF72 is the most common cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). Both the function of C9ORF72 and the mechanism by which the repeat expansion drives neuropathology are unknown. To examine whether C9ORF72 haploinsufficiency induces neurological disease, we created a C9orf72-deficient mouse line. Null mice developed a robust immune phenotype characterized by myeloid expansion, T cell activation, and increased plasma cells. Mice also presented with elevated autoantibodies and evidence of immune-mediated glomerulonephropathy. Collectively, our data suggest that C9orf72 regulates immune homeostasis and an autoimmune response reminiscent of systemic lupus erythematosus (SLE) occurs in its absence. We further imply that haploinsufficiency is unlikely to be the causative factor in C9ALS/FTD pathology.

Functional studies in C. elegans and zebrafish support the haploinsufficiency hypothesis by demonstrating that a reduction in C9ORF72 homolog levels results in locomotion defects 21,22 . However, mouse studies suggest otherwise. Conditional C9orf72 ablation in neurons and glial cells or intracerebral mRNA knockdown did not cause motor neuron disease, gliosis, TDP-43 pathology, or increased ubiquitination, defects associated with C9ALS/ FTD 23,24 . These results imply haploinsufficiency in the central nervous system (CNS) is not pathogenic; however ablation may not have occurred in a crucial cell type and knockdown could have allowed residual C9orf72 expression. The conflicting results and variability intrinsic to cell-specific gene ablation or message knockdown warrant further study of C9orf72 in a universal knockout (C9orf72 −/− ).
In this study, we separate effects intrinsic to the repeat from the function of C9ORF72 by genetically ablating the mouse homolog, 3110043O21Rik (hereafter referred to as C9orf72). Heterozygous (C9orf72 +/− ) mice were healthy and lived normal lifespans. In contrast, C9orf72 −/− mice developed an autoimmune phenotype consisting of expansions in myeloid and lymphoid cell populations, autoantibody production, and glomerulonephropathy. Mild, nonspecific neurological deficits arose after the immune response was established but C9orf72 ablation did not result in classic motor neuron degeneration. Our results indicate haploinsufficiency is not the main cause of C9ALS/FTD neuropathology and describe a novel role for C9ORF72 in immune homeostasis.

Results
C9orf72 −/− mouse engineering and validation. To create a model of C9orf72 haploinsufficiency, we replaced the mouse C9orf72 coding sequence and introns with a lacZ reporter (Supplemental Fig. 1A). To confirm C9orf72 ablation, we performed gene-specific Taqman analyses on wildtype (WT), C9orf72 +/− and C9orf72 −/− tissue cDNA. We detected high C9orf72 expression in WT central nervous system (CNS), fat, and muscle with lower levels in lymphoid tissues. C9orf72 +/− mice had roughly half the expression level of WT, and C9orf72 −/− mice had no detectable C9orf72 expression (Supplemental Fig. 1B). Finally, we confirmed no difference in transcription levels of nearby loci Mob3b, Ak045932, and Ifnk, indicating insertion of lacZ impacts C9orf72 expression only (Supplemental Fig. 1C, data not shown).
Consistent with Taqman results, staining for lacZ in tissues from 6 and 28 week C9orf72 −/− revealed enzyme activity in the brain, spinal cord, spleen, testes, and kidney, corresponding to previously published findings 23,25 . We also observed staining in additional tissues, including fat, muscle, atria, liver, and lung (Supplemental Fig. 1D, data not shown). Reporter activity was more limited in intensity and scope in C9orf72 +/− tissues, as expected for a single lacZ replacement allele.
C9orf72 −/− mice show mild motor deficits. Given the association of C9ORF72 to neuropathology, we performed clinical exams 26 to determine whether loss of C9orf72 causes an ALS-like phenotype. At 40 weeks of age, C9orf72 −/− showed progressive weakness and collapse of hind limbs towards the lateral midline, with mild tremor and rigidity not observed in WT or C9orf72 +/− (Supplemental Fig. 2A, data not shown). Open field observations demonstrated decreased locomotor behaviors and fewer rearing events in null mice compared with WT. CatWalk gait analyses also revealed signs of impaired lower interlimb coordination and reduced stride length with bradykinesia and dragging of hind limbs (Supplemental Fig. 2B). No difference between WT and C9orf72 −/− mice was observed in maximum time spent on the rotarod (Supplemental Fig. 2C).
C9orf72 −/− mice exhibit lymphadenopathy and splenomegaly. During neurological testing, we unexpectedly noted palpable cervical masses in all C9orf72 −/− animals, but not in WT or C9orf72 +/− controls. Masses were palpable as early as 8 weeks of age and present in all null mice by 18 weeks, before onset of observed motor deficits. Upon dissection, the masses proved to originate from cervical lymph node (LN) (Fig. 1A, data not shown) and systemic lymphadenopathy was noted in certain null mice. Peyer's patches (PP) were also enlarged and splenomegaly was apparent by 8 weeks of age (Fig. 1B, data not shown). By 36 weeks, C9orf72 −/− ceased gaining weight compared with WT and only 9 out of 17 survived to the end of the neurological assay period (60 weeks) (Supplemental Fig. 2D, data not shown).

C9orf72 −/− mice display mixed inflammatory infiltrates in multiple organs. The enlargement of
C9orf72 −/− spleens and LN suggests a disease process such as neoplasm or immune dysregulation, an unexpected finding given that ALS/FTD is not linked to such pathology in human patients. To address these possibilities, histopathology was conducted on spleen and LN from 8-60 week old mice. The basic cellular organization of the enlarged C9orf72 −/− LN was preserved, with immunohistochemistry (IHC) identifying a B cell-rich rim (CD45R + ) arranged in follicles within the cortex and a T cell (CD3 + ) infiltrate between follicles and in the paracortex zone (data not shown). A mixed cell population consisting mostly of large round cells with variably distinct borders, a single round nucleus, and eosinophilic, foamy cytoplasm expanded the cortical and medullary nodal architecture. A similar cellular infiltrate was present in C9orf72 −/− spleen, predominantly located within the red pulp, expanding the splenic architecture and corresponding to grossly increased spleen weights. These cells did not stain consistently with CD45R or CD3 but were strongly positive for the macrophage lineage marker F4/80. (Fig. 1C, data not shown). Abundant plasmacytoid cells containing perinuclear halos, consistent with plasma cell morphology, and occasional Mott cells (enlarged plasma cells containing cytoplasmic immunoglobulin (Ig) inclusions) were also present. We did not observe similar infiltrates in WT and C9orf72 +/− controls.
H&E and IHC analyses of additional organs from mice aged 8-60 weeks revealed a prominent F4/80 + population of elongated to angular cells in the liver and kidneys of null mice. This population was pronounced in C9orf72 −/− mouse liver at 8 weeks, though there was no evidence of associated liver disease (data not shown). Increased F4/80 + cell populations observed in C9orf72 −/− kidney were located primarily within the cortex, forming prominent cuffs around glomeruli and aggregates in the vicinity of the macula densa and adjacent tubules (Fig. 1C). Increasing infiltrates of mixed leukocytes were also observed with age, accompanied by varying degrees Scientific RepoRts | 6:23204 | DOI: 10.1038/srep23204 of immune-mediated glomerular disease that was well established by 35-60 weeks. Inflammation was not present in brain or spinal cord tissue in any animals examined. These data implicate the spleen, LN and kidney as major sites of C9orf72 −/− immune pathology.

C9orf72 −/− lymphoid organs contain increased percentages of myeloid lineage cells.
To further interrogate the cellular infiltrate observed on histopathology, flow cytometric analysis was performed on spleen, cervical and mesenteric LN, PP, BM, blood and kidney from C9orf72 −/− and WT controls. Specific focus was on the 28-35 week time point in females, as the majority show renal pathology but remain viable. Total CD45 + (leukocyte common antigen) cell counts were increased in all C9orf72 −/− tissues examined, consistent with the immune infiltration described above, however CD45 + percentages compared with total cell populations assayed were either unchanged or reduced compared with WT. To determine if homeostasis within leukocyte subsets was altered, we narrowed our focus using specific antibody panels. F4/80 + macrophages (CD45 + CD11b + F4/80 + Ly6G − ) were increased in the spleen, LN, kidney, and blood in C9orf72 −/− mice, consistent with the F4/80 + infiltration observed by IHC ( Fig 1D). Neutrophil (CD45 + CD11b + Ly6G + Ly6C int CD115 − ) and total monocyte (CD45 + CD11b + CD115 + ) percentages were also increased in C9orf72 −/− tissues compared with controls. Staining for Ly6G and Ly6C revealed an increased percentage of inflammatory monocytes (CD45 + CD11b + CD115 + Ly6G − Ly6C hi ) in C9orf72 −/− spleen, LN, blood, and kidney. Myeloid dendritic cell (CD45 + CD11b + CD11c + MHCII + ) percentages measured in the spleen, cervical LN, mesenteric LN, PP, BM, and blood were increased in C9orf72 −/− relative to WT, whereas the NK cell (NKp46 + CD49b + ) fraction was decreased in spleen and BM. Co-staining for the activation marker CD86 revealed activated myeloid DC populations (CD45 + CD11b + CD11c + MHCII + CD86 + ) in all tissues examined (data not shown). Similar perturbations in myeloid cell populations by FACS analyses on males aged 9-60 weeks are summarized in Supplemental   Fig. 2A, data not shown). IL-12total was approximately 6-fold increased in C9orf72 −/− animals compared with controls. IL-10, IL-17a, and TNFα were also upregulated, although to a lesser extent. We did not observe changes in IL-1β , IL-2, or IL-4 suggesting this effect on cytokine secretion was not global. While there was a trend toward increased IL-6 in C9orf72 −/− serum compared with WT, this difference did not reach statistical significance. MCP-1 chemokine was significantly increased in female C9orf72 −/− animals and IFNγ was significantly increased in males with an increasing trend observed in females.

RNAseq analyses reveal global inflammatory gene signatures in
C9orf72 −/− mice. C9orf72 ablation appears to cause a systemic immune response resulting in elevated inflammatory cytokines and myeloid cell expansion. High expression of C9orf72 has been observed in monocytes, macrophages and DCs, with lower levels measured in lymphocytes. C9orf72 may indeed modulate the immune system, particularly the myeloid compartment (Supplemental Fig. 5) 27 . To confirm this observation, we performed RNASeq analyses, mapping global transcriptome changes between WT and C9orf72 −/− brain and lymphoid tissues. In the brain, hierarchical clustering primarily separated samples by gender and age, indicating that profiling differences in this tissue are due to the basic biology of samples (data not shown). In contrast, spleen and LN samples clustered based on genotype, with age and sex secondary, thus transcriptome differences in these organs result from loss of C9orf72 expression. Furthermore, over 100 loci associated with immune function showed significant expression differences between C9orf72 −/− and WT tissues in both genders and ages. C9orf72 −/− spleen and LN gene signatures indicate myeloid infiltration with a decrease in the lymphocytic signature. (Fig. 2B). A NextBio enrichment analysis of this sample set revealed the strongest perturbations in gene sets involved in immune response, mouse models of inflammatory conditions, and human infectious diseases, consistent with C9orf72 ablation resulting in global immune dysregulation.

C9orf72 −/− mice have increased percentages of activated T lymphocytes and plasma cells. An increase in activation markers on monocytic cells is often accompanied by increases in lymphocyte activation
parameters. In addition, the elevation in IL-12 observed in C9orf72 −/− serum could up regulate T cell activity 28 . To quantify T cell populations, we performed flow cytometric analyses on spleens, LN, BM, blood and kidney from 30-35 week C9orf72 −/− and WT female mice, gating on CD45 leukocyte common antigen, CD4 or CD8, and various activation status markers. Percentages of CD45 + CD8 + and CD45 + CD4 + cells were reduced overall in C9orf72 −/− animals compared with controls, likely a consequence of myeloid expansion and consistent with RNA profiling results. Conversely, but reflective of expansion of lymphoid tissue and immune infiltration, total cell counts for these T cell populations were increased in null mice (data not shown). Co-staining CD8 + T cells with activation markers revealed increases in the early activation and effector memory T cell markers CD69 and CD44 in C9orf72 −/− spleen and kidney compared with controls. We observed significantly increased percentages of CD8 + T cells expressing PD-1, a co-inhibitory receptor up regulated on activated cells with an important role in down-regulating the immune system. Cervical LN showed significantly increased expression of CD44 and PD-1, although CD69 expression was relatively unchanged (Fig. 3A). CD44, CD69, and PD-1 expression was also increased on CD4 + T cells in C9orf72 −/− spleen, LN, and kidney (Fig. 3B). We also measured less pronounced increases in activated T cell populations in the blood and BM (data not shown). CD4 + FoxP3 + regulatory T (Treg) B cell (CD45 + B220 + CD19 + ) percentages were either unchanged or reduced in C9orf72 −/− spleen but increased in LN compared with controls. Similar to T cell populations, total B cell counts were increased overall in spleen and LN, reflective of lymphoid expansion (data not shown). Analysis of specific B cell subsets revealed significantly increased marginal zone (MZ), (CD45 + CD19 + B220 + CD21 + CD23 − ), follicular (FO) (CD45 + CD19 + B220 + CD21 cervical LN compared with WT. B cell populations in null spleens were overall unchanged or reduced however, similar to myeloid DC, staining for the activation marker CD86 demonstrated an increased activated B cell population (CD45 + CD19 + B220 + CD86 + ) in both C9orf72 −/− spleen and LN (Supplemental Fig. 8A,B). Representative FACS plots are provided in Supplementary Fig. 8C. CD138 co-staining demonstrated increased percentages of mature plasma cells (CD45 + CD19 − B220 − CD138 + ) in C9orf72 −/− LN, spleen, and BM, and an expanded population of B cells transitioning to plasma cells (CD45 + CD19 int B220 int CD138 + ) in all three tissues (Fig. 4A-D and data not shown). We did not find consistent differences between C9orf72 −/− and WT in these cell types in the blood

C9orf72 −/− mice have high titers of autoantibodies. Expansions in plasma cells and transitioning B
cells/plasmablasts can be associated with neoplasms such as multiple myeloma and plasmacytoma 29,30 , in addition to autoimmunity 31,32 . The spleen and LN of C9orf72 −/− animals were enlarged by infiltrates of F4/80 + foamy macrophages occupying appropriate regions for their lineage without obliteration of tissue architecture. Despite showing abnormal proliferation, the mitotic index was low, with only rare mitoses (data not shown). Thus, while a pre-neoplastic condition cannot be excluded, the presence of plasma cells, occasional Mott cells, and glomerulonephritis are more indicative of a potential primary autoimmune process. Serum chemistry panels demonstrated elevated globulin in C9orf72 −/− mice compared with controls and ELISAs showed significantly increased total IgG and IgM in C9orf72 −/− mouse serum (data not shown). We therefore tested WT, C9orf72 +/− , and C9orf72 −/− serums for the common autoantibody, anti-rheumatoid factor (RF), high titers of which are linked to a variety of autoimmune disorders 33 . Both IgG and IgM-type anti-RF titers were significantly elevated in C9orf72 −/− serum compared with controls (Fig. 4E,F).
Aging C9orf72 −/− mice exhibit varying degrees of proliferative glomerulonephropathy. As prefaced earlier, F4/80 + monocytes were present in high numbers in C9orf72 −/− kidneys (Fig. 1C) with evidence of progressive glomerular disease observed by histopathology. To further characterize renal changes in null mice, H&E stained kidney sections were analyzed and scored in five categories of disease. Results showed significantly higher average scoring for membranoproliferative glomerulonephritis in C9orf72 −/− with evidence of occasional glomerulosclerosis, hyaline casts, basophilic tubules and interstitial mononuclear inflammation compared with WT controls (Fig. 5A). Individual histopath scores are shown in Supplemental Table 1. H&E staining of the renal cortex (Supplemental Fig. 9) is representative of mild (middle row) to marked (bottom row) disease progression. Descriptively, mild changes consisted of glomerular enlargement with increased cellularity, and enlargement of Bowman's space. Moderate to severely affected animals also had tubular changes, increased interstitial leukocytic infiltration, thickened capillary walls, and proliferation of visceral (podocytes) and parietal epithelium. In 3/11 null mice, glomerulosclerosis was present, characterized by expansion of the mesangial matrix with acellular, eosinophilic hyaline material, and a variable degree of periglomerular fibrosis. Tubular changes observed in 4/11 C9orf72 −/− kidneys included basophilic tubules (with degeneration/regeneration), cortical and medullary tubular dilatation, hyaline proteinaceous casts, and interstitial infiltrates of mononuclear cells. Consistent with impaired glomerular filtration and correlative to histopathological renal findings, a serum chemistry panel revealed significant elevation of blood urea nitrogen (BUN) and decreased serum albumin in C9orf72 −/− compared with controls (data not shown). Onset of albuminuria is also indicated by urinary albumin to creatinine ratios (ACR) assayed at 14 and 24 week time points. Elevated ACR observed at 24 weeks in null mice is indicative of a progressing renal disease course (Fig. 5B).
Deposition of soluble immune complexes within the glomerular capillaries, followed by complement fixation can cause renal disease such as immune-mediated glomerulonephropathy 34 (GN). High magnification periodic acid schiff (PAS) staining represented in Fig. 5C demonstrated regional thickening of the glomerular basement membrane in capillary loops associated with immune-mediated GN in C9orf72 −/− mice, but not WT. To further evaluate whether increased total Ig and autoantibody levels contribute to GN, we performed IHC on C9orf72 −/− and WT, 8-63 week kidney sections for total IgG and IgM as well as complement C3 on a 60-63 week surviving cohort of females. Overall, C9orf72 −/− kidneys demonstrated increased IgG immunostaining compared with WT at all time points, with diffuse, intense IHC signal in the vasculature and tubular epithelium of the medulla and cortex observed as early as 8 weeks. Correlating with the onset of GN pathology, we noted an increase in glomerular IgG and IgM staining by 38 weeks (Fig. 5C, Supplemental Fig. 9). Staining for both IgG and IgM was also frequently associated with the parietal layer of Bowman's capsule. IgG was occasionally observed in the urinary space and/or within proximal renal tubules, indicating impaired glomerular filtration. Intense IgG staining was present in tubule epithelial cells in animals with severe disease, consistent with reabsorption of abundant IgG. Occasionally, similar, less intense staining was observed for IgM. Staining in sclerotic glomeruli was diminished compared with WT, consistent with impaired blood flow to these units. However, glomeruli that retained patent vascular loops tended to have increased IgG in comparison with WT. Fine granular deposits and/or linear staining of IgG and IgM associated with vascular membranes suggestive of immune complex deposition were frequently present (Fig. 5C). Complement factor C3 deposition is commonly associated with immunoglobulin deposits on basement membranes in immune-mediated glomerular disease 34 . IHC for C3 revealed increased staining in glomerular tufts of C9orf72 −/− mice compared with WT (Supplemental Fig. 9A). Granular and linear staining was most prominent on the membranes of the visceral layer of glomerular capsule, delineating the capillary loops and podocytes as depicted in 60× magnification (Fig. 5C).

C9orf72 −/− mice develop SLE-like disease. Systemic lupus erythematosis (SLE) is characterized by
immune dysregulation affecting many organs of the body 35 . C9orf72 −/− mice develop the lymphoid hyperplasia, anemia, and renal disease common in SLE patients and reminiscent of phenotypes observed in spontaneous mouse models of SLE such as the MRL/lpr and NZB/W F1 strains 36 . A hallmark of SLE is high titer antinuclear antibodies (ANA) of certain specific types 37 . We therefore tested C9orf72 −/− and control mouse serum for ANA encompassing a subset of autoantibodies against proteins and structures in the nucleus; specifically, we tested for anti-Smith (Sm) antibodies that recognize core units of small nuclear ribonucleic proteins (snRNP), anti-double-stranded DNA (dsDNA) antibodies, as well as anti-cardiolipin antibodies that are reactive to an essential element of the inner mitochondrial membrane. ANA, anti-dsDNA and anti-cardiolipin were significantly Scientific RepoRts | 6:23204 | DOI: 10.1038/srep23204 increased in female C9orf72 −/− mice by 8 weeks compared with controls. Anti-Sm was slightly elevated at 8 weeks in null mice with a dramatic increase above WT observed by the 24 week timepoint (Fig. 6A, data not shown).
Increased autoantibody titers in lupus patients are positively correlated with an increased frequency of circulating T follicular helper (Tfh) cells 38 . Interrogation of this specific cell population (CD4 + CXCR5 + CD44 + ICOS + PD-1 + Bcl-6 + ) in spleen, cervical LN, mesenteric LN, and blood by FACS analysis revealed significantly increased Tfh cell populations in C9orf72 −/− tissues compared with controls. (Fig. 6B data not shown). Elevated Tfh cells were week timepoints from the same cohort of mice indicate onset of albuminuria in C9orf72 −/− animals with age. Heterozygous mice display values comparable to WT consistent with the absence of an observed phenotype. (C) PAS Staining and IHC immunoreactivity of mouse glomeruli for IgG, IgM, and C3 demonstrate immunemediated kidney damage. Mesangial matrix (red chevron) is PAS+ and markedly expanded in C9orf72 −/− mouse compared with the delicate matrix seen in the WT. Basement membrane of the vasculature is markedly expanded by PAS+ matrix (black arrows), obliterating the capillary loop. The parietal epithelium is proliferative (*), and Bowman's capsule is surrounded by connective tissue and mononuclear cells (black chevron). IHC for IgG, IgM, and C3 in WT (top panel) shows delicate capillary loops (black arrows) supported by scant mesangial matrix; faint chromogenic staining of the capillary endothelia is representative of physiologic IgG, IgM, and C3. IHC stippling is also observed in the mesangium (red chevrons). In contrast, glomeruli from C9orf72 −/− mice have increased mesangial matrix (red chevrons) and granular deposits of IgG, IgM, and C3. Capillary loop basement membranes are thickened and delineated by subendothelial/subepithelial granular to confluent deposits (black arrows). Chromogenic staining for IgG is also apparent in surrounding renal tissue. Note the increased size of the C9orf72 −/− glomerulus compared with that of the WT, and the increase in the urinary space of the glomerulus. also observed in C9orf72 −/− BM that did not reach significance (data not shown). Collectively, our data suggest an immune response similar to human SLE occurs in the absence of C9orf72 expression.

Discussion
The association between C9ORF72 GGGGCC repeat expansion and neurological disease is well-established, but the pathogenic mechanism remains elusive. The absence of phenotype in C9orf72 +/− mice presented in our study contradicts haploinsufficiency of C9ORF72 as the main cause of C9ALS/FTD pathology. More importantly, our study highlights a novel immune regulatory role for C9orf72 through the first comprehensive phenotypic analysis of a mouse line with global C9orf72 ablation. Since analog immunological findings are not typically present among ALS patients, our data indicate that C9orf72 gene function is unrelated to known C9ALS/FTD pathology, making a nonspecific effect of the repeat expansion on C9ALS/FTD pathology more likely.
Global ablation of C9orf72 resulted in select expansions of myeloid and lymphoid compartments, with increased T, B and DC cell activation and elevated plasma cells. C9orf72 −/− mice demonstrated elevated serum IL-12 and other cytokines, in addition to tissue RNA signatures consistent with myeloid upregulation. Renal disease with accompanying pathological changes was present in the majority of mice by 35 weeks. At a microscopic level, glomeruli stained heavily with antibody to Ig and C3 in a pattern suggesting immune complex deposition. Null mice also had increased Tfh cell populations and high titer anti-RF, ANA, anti-Sm, and anti-cardiolipin autoantibodies that are commonly associated with human SLE 37,38 . In summary, loss of C9orf72 expression profoundly disturbs immune homeostasis and results in systemic autoimmune disease.
SLE is a constellation of immunological abnormalities affecting multiple organ systems. Patients may have manifestations in the kidney, skin, joints, lungs, and/or heart that usually follow a fluctuating course, with periodic flares and periods of reduced disease activity 35 . 95-98% of patients have elevated ANA and an elevated anti-Sm antibody titer is pathognomonic for SLE 39 . Mouse models of SLE include the spontaneously occurring MRL/lpr and NZB/W F1 strains, as well as other inducible models, however none fully represent human disease 36 . NZB/W F1 mice replicate the female bias toward worsened pathology, elevated ANA with anti-dsDNA and immune complex-mediated glomerulonephritis, but do not develop anti-Sm ribonuclear protein antibodies. MRL/lpr mice have inflammation in many classic SLE sites and elevated anti-Sm and anti-cardiolipin, although males and females are equally affected. Neither of these strains or any single-gene or inducible model replicates the waxing and waning in symptoms characteristic of human disease. C9orf72 −/− animals display a subset of lupus-like symptoms, namely lymphoid activation and hyperplasia, a characteristic autoantibody profile and evidence of GN. They do not replicate all aspects of human SLE and the same caveats relevant to established models should also be applied to our mice.
Human SLE may also comprise neuropsychiatric symptoms, including cognitive dysfunction, mood disorders, psychosis, and cerebrovascular diseases, collectively referred to as neuropsychiatric SLE (NP-SLE). Onset of symptoms may be associated with high titer autoantibodies, increased blood-brain barrier (BBB) permeability, and cytokine production 40 . The MRL/lpr mouse strain develops behavioral abnormalities and is commonly used as a model to study such manifestations 41 . High titer anti-cardiolipin correlates positively with neuropsychiatric symptoms 42,43 and is observed in both MRL/lpr and our C9orf72 −/− mice. We observed mild motor impairment in C9orf72 −/− that suggested neurological deficit, however lymphoid organ hyperplasia and GN were already established, suggesting defects could be secondary to the immune phenotype. We did not observe overt pathology or marked transcriptome changes in C9orf72 −/− brain and spinal cord, however we cannot rule out highly localized disease processes not obvious by our assays. Given the growing body of evidence supporting a link between autoimmune mechanisms and neurological disease 44 , studies to evaluate behavioral and cognitive disorders in C9orf72 −/− mice should also be considered.
C9orf72 expression profiling in immune cell types assembled by the Immunological Genome Project (Immgen) and additional microarray expression studies reflect high expression of C9orf72 in monocyte, macrophages and DC populations 27,45 . Relative to this insight, expanded proliferative potential of the monocyte system with increased F4/80 + macrophages has previously been described in MRL/lpr and NZB/W mice. These strains show early involvement of the BM in monocytopoiesis, and retain up-regulated extramedullary monocyte proliferation, in contrast to control strains that down-regulate this process as the BM develops its full activity 46 . Defects in phagocytosis intrinsic to F4/80 + cells have been described in mouse SLE models and human patients 47,48 and inefficient clearance of apoptotic debris is considered a hallmark of SLE 49 . Indeed, polymorphisms in the autophagy-related gene, ATG5 are linked to increased susceptibility to SLE 50,51 , as is activation of the negative regulator of autophagy, mTOR 52 . Alternatively, other studies have reported enhanced autophagy in patient and mouse model B cells 53,54 and patient serum factors are capable of inducing autophagy in neuroblastoma cell lines 55 . We have not yet assessed the phagocytic potential of the predominating F4/80 + cells in C9orf72 −/− mice, but such experiments could determine whether dysregulation of autophagy is associated with C9orf72 ablation and contributing to the SLE phenotype or is simply a byproduct of other processes.
C9ORF72 protein is widely expressed and highly conserved as a single-gene copy across all vertebrates with remarkably high sequence identity across species. Notably, it has not been duplicated, even in fish genome, all of which have undergone rounds of duplication suggesting its function is crucial, but under constraint. Although still largely unknown, C9ORF72 function has been linked to intracellular trafficking via Rab-dependent pathways essential for endosomal transport. Depletion of C9ORF72 by siRNA in neuronal cell lines inhibited endocytosis and dysregulated autophagy, an important process for cellular homeostasis 56 . Given that autophagy can protect against neurodegenerative disease by preventing accumulation of toxic proteins, disrupting this process by reducing C9ORF72 expression may render cells more susceptible to repetitive RNA and the products of erroneous hexanucleotide repeat translation.
Recent studies using human C9ORF72 isoform-specific antibodies demonstrated colocalization of the short isoform with components of the nuclear pore complex. Interestingly, motor neurons derived from C9ALS/ FTD patients exhibited loss of the short C9ORF72 isoform and mislocalization of TDP-43, indicating defects in nucleocytoplasmic shuttling 57 . More recently, however, three studies suggest the repeat expansion itself can affect shuttling independent of its surrounding genetic locus. Yeast expressing dipeptide inclusions and Drosophilia expressing RNA repeats presented with defects in intracellular transport, a finding then replicated in repeat-transfected cell lines, patient inducible pluripotent stem cell neurons and C9ALS-FTD patient brain tissue [58][59][60] . To date, all relevant protein function and repeat-associated studies have been restricted to neuronal cell lines and tissue samples. Further experimentation in a wider array of cell types could help to elucidate a specific role for C9ORF72 and separate processes relevant to protein function from those associated with the repeat. Furthermore, pursuing similar experiments outside the CNS could reveal novel relationships between autoimmunity and neurodegeneration and unveil new pathways with therapeutic potential.
In summary, our results implicate loss of mouse C9orf72 expression with autoimmunity that resembles human SLE and suggest a new role for C9ORF72 as an important regulator of the immune system. C9ORF72 has not yet been linked to SLE in humans and pursuit of potential disease variants with a specific focus on autoimmune populations should be considered. Further experimentation in cell lines outside the CNS should also be undertaken to help elucidate the function of C9ORF72. In addition, we propose loss of C9orf72 protein is not the likely cause of C9ALS/FTD neuropathology. Generation of a mouse model with the repeat expansion targeted into the C9orf72 locus to more closely mimic the human genetic lesion will help to address C9ALS/FTD pathological mechanisms.

Methods
Generation of C9orf72 −/− mice. We employed the VelociGene ® and VelociMouse ® methods as described previously [61][62][63][64] in which targeted ES cells (F1 hybrid 129S6SvEvTac/C57BL6NTac ES cells) were injected into uncompacted 8-cell stage Swiss Webster embryos to produce healthy fully ES cell-derived F0 generation mice heterozygous (Het) for the C9orf72 mutation. F0 generation male Hets were crossed with C57Bl6/NTac females to generate F1 Hets that were intercrossed to produce F2 generation WT, C9orf72 +/− and C9orf72 −/− mice for phenotypic analyses. A second cohort of N2F2 generation mice was generated via in-vitro fertilization (IVF) using frozen F1 heterozygous sperm and oocytes from C57Bl6/NTac donor females. N2F1 Het offspring were then intercrossed to generate N2F2 WT, C9orf72 +/− and C9orf72 −/− mice for phenotypic analysis. Animals. Phenotypic studies of F2 and N2F2 mice began at 6 weeks of age. Mice were observed from birth for various developmental milestones (runting, breathing, facial and limb abnormalities, skin color, posture, righting and eye opening) until 6 weeks of age, when they were housed 2-5 per cage in 12 hours of light per day at 20-23 °C, and 40-60% humidity for study. Mice were housed in 95.6 × 309.1 × 133.4 mm cages (Thoren) with cob bedding (The Andersons Lab Bedding) and a cotton nestlet for enrichment (Ancare). Mice had access to normal chow (LabDiet) and water ad libitum and were monitored twice daily for health status. All animal procedures were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and were approved by the Regeneron Pharmaceuticals Institutional Animal Care and Use Committee (IACUC).
Taqman expression analysis. Axillary and brachial lymph nodes, cervical lymph nodes, gonadal fat pad, frontal cortex, diaphragm, spinal cord, spleen, and thymus were dissected fresh into RNALater stabilization reagent (Qiagen) and stored at − 20 °C. Tissues were homogenized in TRIzol and chloroform was used for phase separation. The aqueous phase, containing total RNA, was purified using miRNeasy Mini Kit (Qiagen, Cat#217004) according to manufacturer's specifications. Genomic DNA was removed using MagMAX ™ Turbo ™ DNase Buffer and TURBO DNase (Ambion by Life Technologies). mRNA was reverse-transcribed into cDNA using SuperScript ® VILO ™ Master Mix (Invitrogen by Life Technologies, Cat# 11755500). cDNA was amplified with the TaqMan ® Gene Expression Master Mix (Applied Biosystems by Life Technologies, Cat# 4370074) using the ABI 7900HT Sequence Detection System (Applied Biosystems). Actb was used as the internal control gene to normalize cDNA input differences. WT thymus was used as a reference sample to calculate the fold difference of mRNA between samples.
Behavioral scoring tests. The assessment of overall motor function was performed using blinded subjective scoring assays. Motor impairment score was measured using the system generated by ALS Therapy Development Institute (ALS TDI) 26 . Tremor and rigidity scores were measured using a 0-3 scale, where 0 = no symptoms, 1 = mild, 2 = moderate, and 3 = severe. Locomotor behaviors were evaluated for 60 minutes every other week using the automated Open Field system (Kinder Scientific), which measures fine movements, X+ Y ambulation, distance traveled, number of rearing events, time spent rearing, and immobility time via infrared beam breaks. Rotarod (IITC Life Science, Woodland Hills, CA) was used to measure the latency for a mouse to fall from a rotating beam with a ramping speed, starting at 1 rpm and accelerating to 15 rpm over 180 seconds. The average and maximum of the three longest durations of time that the animals stay on the beam without falling were used to evaluate latency. Gait analysis was performed using the CatWalk XT 10 (Noldus). Mice spontaneously ambulated across a runway and the footprints were automatically analyzed for interlimb coordination (the percentage of normal step sequences) and stride length (the distance between successive placements of the same paw).
Histology. Tissues were directly harvested into 4% PFA or collected following transcardial perfusion with 50 mL of saline solution, 50 mL of 4% PFA solution in acetate buffer at pH 6.5 and finally 50 ml of 4% PFA solution in borate buffer at pH 9.5. Spinal cords were collected into 15% followed by 30% sucrose solution in borate buffer until they sank. All other tissues were post-fixed in 4% PFA and transferred to 70% ethanol after 24 hours. Paraffin embedding, sectioning, and H&E staining were performed by Histoserv, Inc. (Germantown, MD). PAS staining and IHC for IgG, IgM, C3, CD45R, CD3, CD138, and F4/80 was completed by Histotox Labs, (Boulder, CO). Histopathological scoring of H&E stained kidney sections was performed by a blinded, board certified veterinary pathologist and evaluated in disease categories according to guidelines set forth by The International Harmonization of Nomenclature and Diagnostic Criteria for Lesions in Rats and Mice (INHAND) Project 65 .