Abstract
Amyotrophic lateral sclerosis is a fatal neurodegenerative disease with paralysis resulting from dysfunction and loss of motor neurons. A common neuropathological finding is attrition of motor neuron dendrites, which make central connections vital to motor control. The chromatin remodelling complex, neuronal Brahma-related gene 1 (Brg1)-associated factor complex (nBAF), is critical for neuronal differentiation, dendritic extension and synaptic function. We have identified loss of the crucial nBAF subunits Brg1, Brg1-associated factor 53b and calcium responsive transactivator in cultured motor neurons expressing FUS or TAR-DNA Binding Protein 43 (TDP-43) mutants linked to familial ALS. When plasmids encoding wild-type or mutant human FUS or TDP-43 were expressed in motor neurons of dissociated spinal cord cultures prepared from E13 mice, mutant proteins in particular accumulated in the cytoplasm. Immunolabelling of nBAF subunits was reduced in proportion to loss of nuclear FUS or TDP-43 and depletion of Brg1 was associated with nuclear retention of Brg1 mRNA. Dendritic attrition (loss of intermediate and terminal dendritic branches) occurred in motor neurons expressing mutant, but not wild-type, FUS or TDP-43. This attrition was delayed by ectopic over-expression of Brg1 and was reproduced by inhibiting Brg1 activity either through genetic manipulation or treatment with the chemical inhibitor, (E)-1-(2-Hydroxyphenyl)-3-((1R, 4R)-5-(pyridin-2-yl)-2, 5-diazabicyclo[2.2.1]heptan-2-yl)prop-2-en-1-one, demonstrating the importance of Brg1 to maintenance of dendritic architecture. Loss of nBAF subunits was also documented in spinal motor neurons in autopsy tissue from familial amyotrophic sclerosis (chromosome 9 open reading frame 72 with G4C2 nucleotide expansion) and from sporadic cases with no identified mutation, pointing to dysfunction of nBAF chromatin remodelling in multiple forms of ALS.
Introduction
Amyotrophic lateral sclerosis (ALS) is an invariably fatal adult onset neurodegenerative disease in which paralysis, and eventually loss of respiratory function, results from dysfunction and loss of cortical, brainstem and spinal motor neurons controlling muscle movement.
ALS is a syndrome with multiple causes including mutations in several genes underlying familial forms of the disease. Among ALS-linked genes are those encoding several heterogenous ribonucleoproteins (hnRNPs) with DNA/RNA-binding activities involved in transcription, RNA splicing and mRNA transport including FUS, encoding fused in sarcoma/translated in liposarcoma (FUS) (ALS6) (1,2) and TARDBP encoding TAR-DNA Binding Protein 43 (TDP-43) (ALS10) (3–5). Although, these proteins undergo nucleocytoplasmic shuttling to fulfil some of their many roles in regulating RNA metabolism, they normally are detected predominantly in the nucleus; however, nuclear depletion and accumulation in neuronal cytoplasmic inclusions have been observed in autopsy tissue from cases with mutations in these genes as well as in cases with other forms of familial and sporadic ALS (6–8), pointing to a common pathway in multiple types of ALS.
Attrition of motor neuron dendrites is a common, but poorly understood, neuropathological feature of ALS (9–11). FUS and TDP-43 are important for determining dendritic morphology. FUS is involved in transport of mRNAs for local synthesis at dendritic spines in response to neuronal activity, with FUS knockout resulting in abnormal spine morphology and dendritic branching (12). TARDBP-null Drosophila neurons exhibited defective dendritic branching, which was rescued by re-introducing wild-type (WT) TDP-43, but not ALS-associated TDP-43 mutants (13). In addition, cortical neurons and spinal motor neurons in mutant FUS transgenic mice displayed reduced dendritic complexity (14,15). Dendritic attrition is a prominent feature in cultured murine motor neurons expressing ALS-linked mutants including FUS (16) and TDP-43 (this study), establishing models in which to investigate mechanisms and potential interventions.
Neuronal Brg1/Brm Associated Factor (nBAF) chromatin remodelling complexes (CRC) play an important role in regulating gene expression by remodelling nucleosomes and thereby accessibility of transcription factors to gene promoter elements. CRC are vital to regulation of gene expression and the formation of nBAF complexes is a key requirement for neuronal differentiation including extension of processes (17). Our interest in nBAF CRC in ALS arose from this role in establishing neuronal architecture and in remodelling of dendritic connections in response to neuronal activity (18,19). In addition, de novo mutations in calcium-response transactivator (CREST) have been identified by exome sequencing of ALS trios (ALS patients and both unaffected parents) (20) and FUS interacts with nBAF proteins (20).
BAF CRC in vertebrates are composed of at least 15 subunits including an ATPase/DNA helicase [Brg1/Swi/Snf Related Matrix Associated Actin Dependent Regulator of Chromatin, Subfamily A 4 (SMARCA4) or Brm/SMARCA2] (21). Particular BAF complexes are associated with embryonic stem cells [embryonic stem cell Brg1 associated factor (esBAF)], neural progenitors (npBAF) and postmitotic neurons (nBAF). Conversion of npBAF to nBAF complexes requires substitution of key subunits: BAF53a is replaced by Brg1-associated factor 53b (BAF53b); BAF45a/d is replaced by BAF45b/c; and synovial sarcoma translocation, chromosome 18 (SS18) is replaced by CREST.
Using our primary culture model of ALS6, in which WT human FUS or ALS-linked mutants are expressed in motor neurons of dissociated murine spinal cord cultures (16,22), we identified loss of nuclear Brg1 and other nBAF components when FUS redistributes to the cytoplasm and linked this epigenetic mechanism to the accompanying dendritic attrition. Loss of nBAF subunits also occurred in cultured motor neurons expressing mutant TDP-43 associated with ALS10 and, of particular importance, in spinal motor neurons of ALS autopsy specimens. These data point to dysregulation of chromatin remodelling as an important mechanism of neuronal dysfunction across multiple forms of ALS.
Results
Dendritic attrition in motor neurons expressing mutant FUS
For these studies, we used a previously developed primary culture model of ALS6 in which human WT or ALS-linked mutant FUS is expressed in motor neurons of 3- to 7-week-old murine dissociated spinal cord-dorsal root ganglion cultures by intranuclear microinjection of plasmid expression vectors (16,22). mCherry was coexpressed to visualize cell morphology and to quantify dendritic architecture by Sholl analysis as described in Materials and Methods. Dendritic branching was significantly reduced in neurons expressing mutant FUS compared with WT FUS or empty vector (injection control) (Fig. 1A–C). Root (primary) branches closest to the cell soma were spared. Rather, intermediate and terminal branches were reduced (Fig. 1C), as well as the average number of processes per neuron and total cable length, a measure of dendritic output (Supplementary Material, Fig. S1A).
Dendritic attrition in motor neurons expressing mutant FUS is linked to loss of nuclear Brg1 expression. (A) Loss of intermediate and terminal dendritic branches in motor neurons expressing mutant FUS. Shown are mCherry epifluorescence images of cultured motor neurons 3 days following intranuclear microinjection of plasmids encoding mCherry along with pcDNA3 empty vector, human WT FUS or the FUS mutant R521G. (B) Sholl analysis comparing dendritic branching of neurons under the conditions described in (A), n = 3 cultures per condition. ‘SD’ indicates a significant difference (P < 0.05. Welch’s t-test) between mutant FUS and empty vector control. (C) Categorization of Sholl curves shown in (B), into Root, Intermediate and Terminal segments. ‘SD’ indicates a significant difference (P < 0.05, Welch’s t-test) between mutant FUS and empty vector control. Further details of Sholl analysis are presented in Supplementary Material, Figure 1SA. (D) Double immunolabelling of neurons expressing Flag-tagged WT or mutant FUS with anti-Flag/anti-Brg1 antibodies on Day 3 post-injection showing examples of the different distribution of FUS. Arrowheads point to nuclei depleted of Brg1. Scale bar = 20 µm. (E) Plotted are fluorescence intensity measurements of nuclear anti-Brg1 antibody labelling in motor neurons expressing WT or R521G FUS according to whether FUS was predominantly nuclear, predominantly cytoplasmic or distributed in both compartments (as illustrated in D). Presented are means ± S.E.M. of background subtracted fluorescence units. The number above each bar is the percentage of motor neurons exhibiting the represented FUS localization. The symbol //designates no cells with that phenotype. Asterisks indicate significant difference compared with neurons expressing WT FUS; *P < 0.05, **P < 0.01, ***P < 0.001 (two-tailed t-test).
Nuclear Brg1 was severely diminished in neurons with cytoplasmic accumulation of FUS and is required to maintain dendritic morphology
Since nBAF complex subunits are important for dendritic outgrowth and branching (17) and interact with FUS (20), we performed double-label immunocytochemistry to investigate whether these subunits were mislocalized along with cytoplasmic accumulation/nuclear depletion of FUS. FUS was detected by anti-Flag, previous studies having shown the same distribution of ectopic FUS and total FUS including the endogenous protein (16). By Day 3 following plasmid-injection, mutant FUS had accumulated in the cytoplasmic in motor neurons to a much greater extent than WT FUS, as in our previous studies (16,22). The localization of FUS was classified as predominantly nuclear, predominantly cytoplasmic or distributed in both compartments (Fig. 1D). The percentage of neurons in each category is presented as the numbers on each bar in the graph in Figure 1E. Overall, the level of Brg1, the main DNA-helicase/ATPase of nBAF CRC, in the nucleus decreased with redistribution of FUS from the nucleus to the cytoplasm, as indicated by reduced fluorescence intensity of anti-Brg1 antibody labelling in motor neuronal nuclei (Fig. 1E). In this figure, fluorescence intensities of Brg1 labelling were binned and plotted according to whether FUS was only nuclear, both nuclear and cytoplasmic or mostly cytoplasmic; note the decrease in mean Brg1 intensity with loss of nuclear FUS.
In order to test whether Brg1 is required for maintenance of dendritic morphology in motor neurons, we examined dendritic architecture after knockdown of Brg1 function, either through pharmacological inhibition or genetic manipulation. Treatment with the inhibitor of Brg1 activity, (E)-1-(2-Hydroxyphenyl)-3-((1R, 4R)-5-(pyridin-2-yl)-2, 5-diazabicyclo[2.2.1]heptan-2-yl)prop-2-en-1-one (PFI-3) (23) for 72 h induced a dose-dependent reduction in dendritic branching (Fig. 2A and C) and Supplementary Material, Fig. S1B), without depleting Brg1 from the nuclei (data not shown). Neurons, expressing either shRNA specific for Brg1 (24), which knocked down Brg1 immunoreactivity by 60–70%, or a dominant-negative mutant of Brg1 (K798R) (25) showed significant dendritic attrition compared with control neurons similar to that induced by expression of mutant FUS. Thus, Brg1 activity is required to maintain dendritic branching in motor neurons and loss of nuclear Brg1 is likely to contribute to the dendritic attrition in motor neurons with depleted nuclear FUS. In support of that conclusion, over-expression of human Brg1 was sufficient to maintain nuclear Brg1 and dendritic architecture in motor neurons expressing FUSR521G, as evaluated on Day 3 post-microinjection of plasmids (Fig. 2B and E and Supplementary Material, Fig. S2). These data indicate that nuclear Brg1 is necessary to preserve established dendritic architecture.
Disrupting Brg1 function results in dendritic attrition and ectopic expression of Brg1 delays dendritic attrition caused by mutant FUS. Representative epifluorescence images of neuronal morphology after 3 days of expressing (A) mCherry either alone or in the presence of the Brg1 inhibitor PFI-3 or Brg1 shRNA, illustrating attrition of dendritic branches by disrupting Brg1; (B) mCherry + empty plasmid, FUSR521G or FUSR521G + Brg1 plasmid (pBJ5Brg1), the latter showing robust dendritic branching. Scale bar = 30 µm. (C–E) Sholl curves of neurons expressing mCherry and (C) treated with vehicle or varying concentrations of PFI-3; (D) expressing either Brg1 shRNA, scramble shRNA, or dominant-negative Brg1 (Brg1 DN) or (E) expressing FUSR521G ± pBJ5Brg1. Presented are means ± S.E.M. of the data. Number of neurons measured for each graph is listed in the graph keys. ‘SD’ indicates significant difference compared with vehicle treated neurons or to mCherry/pcDNA3 empty vector (P < 0.05, Welch’s t-test). Additional details of Sholl analyses are presented in Supplementary Material, Figure S2.
Multiple nBAF subunits were diminished in motor neurons with cytoplasmic FUS
We next investigated whether other subunits of the nBAF complex were dysregulated in addition to Brg1, focusing on CREST and BAF53b. Both proteins are neuron-specific and necessary for neuronal differentiation and activity-dependent dendritic outgrowth, for calcium-dependent regulation of gene expression and directing the CRC to gene promoters, respectively (17,26). Fluorescence intensity measurements of nuclear antibody labelling indicated a significant decrease in nuclear levels of both BAF53b and CREST with nuclear depletion/cytoplasmic accumulation of either WT or mutant FUS (Fig. 3).
The critical nBAF subunits, BAF53b and CREST, are also reduced in motor neurons expressing mutant FUS. (A and B) Double immunolabelling of neurons expressing WT or mutant FUS with antibodies to Flag plus either anti-BAF53b or anti-CREST; Day 3 post-microinjection of plasmid vectors. Arrow heads point to nuclei. Scale bar = 20 µm. (C and D) Fluorescence intensity of anti-BAF53b or anti-CREST antibody labelling of the nuclei of neurons expressing WT or mutant (R521G) human FUS on Day 3 post-injection. The symbol //designates no cells with that phenotype. Asterisks indicate significant difference compared with neurons expressing WT FUS (*P < 0.05, **P < 0.01, ***P < 0.001; two-tailed t-test). The number above each bar is the percentage of motor neurons exhibiting the represented FUS localization.
Nuclear retention of Brg1 mRNA in motor neurons with cytoplasmic retention of FUS
Brg1 mRNA levels were examined by fluorescence in situ hybridization (FISH) in combination with immunocytochemistry to identify the distribution of FUS in the same neurons (Fig. 4). In injected neurons in which WT or mutant FUS was restricted to the nucleus, Brg1 mRNA had a punctate distribution throughout the nucleus and cytoplasm (Fig. 4A); however, in 100% of neurons with cytoplasmic inclusions, Brg1 mRNA was concentrated in the nucleus. In this experiment, FUSR521H showed nuclear only distribution in 17% of motor neurons on Day 3, with the remaining exhibiting significant cytoplasmic inclusions. All neurons with cytoplasmic FUS inclusions had nuclear Brg1 mRNA. No signal was detected using the control sense probe. Thus, the major finding was nuclear retention of Brg1 mRNA in neurons with cytoplasmic FUS inclusions. Triple labelling of neurons for the presence of FUS, Brg1 mRNA and Brg1 protein showed that Brg1 mRNA was nuclear in motor neurons with depleted Brg1 protein and cytoplasmic FUS (Fig. 4C). Quantitation of total mRNA levels/neuron was not feasible given the very different fluorescence intensity of labelling and volume of distribution between nuclear and cytoplasmic mRNA (extending from juxtanuclear region into the dendritic branches).
Brg1 RNA was concentrated in the nucleus of motor neurons with FUS cytoplasmic inclusions. Motor neurons were microinjected with plasmid encoding WT or mutant (R521H or P525L) FUS. After 3 or 6 days, cultures were fixed for combined FISH and immunocytochemistry to visualize Brg1 mRNA and the distribution of FUS in the same neurons, as described in Materials and Methods. (A) On Day 3, Brg1 mRNA was localized to cytoplasmic puncta when WT or mutant FUS was nuclear. In all neurons with FUS cytoplasmic inclusions, Brg1 mRNA localization was shifted to the nucleus. (B) Lack of labelling by the control probe (lower panel) in neurons expressing WT or mutant FUS (upper panel). (C and D) In some experiments, cultures were triple labeled to visualize distribution of Brg1 protein (using rabbit anti-Brg1 and anti-rabbit IgG-Cy5) in the same neurons as Brg1 mRNA and FUS. (C) Nuclear retention of Brg1 mRNA corresponded with reduction in nuclear Brg1 protein and presence of cytoplasmic FUS inclusions (with or without nuclear FUS). (D) On Day 6, neurons with cytoplasmic inclusions of FUSWT could be detected and exhibited nuclear Brg1 mRNA as well as reduced expression of Brg1 protein. Arrows points to nuclei of motor neurons in the images. Scale bar = 20 µm.
Nuclear retention of Brg1 mRNA was not restricted to neurons expressing mutant FUS, also occurring in neurons with cytoplasmic inclusions of WT FUS; however, this occurs in fewer neurons than those expressing mutant FUS, in our studies and in others, and was more evident on Day 6 (Fig. 4D). Thus, with redistribution of FUS, whether WT or mutant, to the cytoplasm, RNA encoding Brg1 is retained in the nucleus, pointing to an mRNA processing/transport defect as the predominant mechanism underlying the reduced Brg1 protein rather than primary effects on protein synthesis or turnover.
Dendritic attrition and nBAF subunit depletion in neurons expressing mutant TDP-43
Dendritic attrition and depletion of nBAF subunits were not unique to FUS. Similar decrease in dendritic branching was measured in motor neurons expressing mutant human TDP-43, compared with neurons expressing WT TDP-43 or control neurons expressing mCherry (Fig. 5A); the decrease in dendritic complexity was owing to loss of intermediate and terminal branches (Supplementary Material, Fig. S3). As in neurons expressing mutant FUS, co-expression of Brg1 delayed mutant TDP-43-induced dendritic attrition (Fig. 5B).
Dendritic attrition and reduction of nBAF subunits in motor neurons expressing mutant TDP-43. (A) Dendritic attrition in motor neurons expressing an ALS-linked mutant TDP-43 revealed by Sholl analysis of dendritic morphology of neurons co-expressing mCherry to visualize morphology plus either WT or mutant (G348C) human TDP-43 for 3 days. The symbol //designates no cells with that phenotype. Asterisks indicate significant difference compared with control (mCherry+empty pcDNA plasmid) (***P<0.001; Welch’s t-test). Reduced dendritic arborization in motor neurons expressing mutant TDP-43 was similar to neurons expressing mutant FUS. (B) Sholl analysis of dendritic morphology of motor neurons showing delay in loss of dendritic processes with ectopic expression of Brg1 along with myc/Flag-tagged TDP-43G348C. ‘SD’ indicates significant difference from mCherry control (P<0.05, Welch’s t-test). (C and D) Diminished expression of multiple nBAF subunits in motor neurons expressing mutant TDP-43. (C) Double immunolabelling of motor neurons expressing WT or mutant (G348C) TDP-43 with anti-Flag and anti-Brg1, anti-BAF53b or anti-CREST on Day 3 post-microinjection of plasmid. Arrowheads point to nuclei of motor neurons. Scale bar=20 µm. (D) Plotted are fluorescence intensity measurements of nuclear nBAF subunit antibody labelling in motor neurons expressing WT or G348C TDP-43, according to whether TDP-43 was predominantly nuclear, predominantly cytoplasmic or distributed in both compartments. Presented are means±S.E.M. The number above each bar is the percentage of motor neurons exhibiting the represented FUS localization. Asterisks indicate significant difference compared with nuclear WT TDP-43 (*P<0.05, **P <0.01, ***P<0.001; two-tailed t-test).
Motor neurons expressing WT or mutant human TDP-43 also showed a decrease in nuclear Brg1, BAF53b and CREST when either WT or mutant protein shifted to neuronal cytoplasm inclusions, as indicated by decreased nuclear antibody fluorescence intensity 3 days post-microinjection of plasmid (Fig. 5C and D). These data point to nBAF subunit depletion as a shared pathway in FUS and TDP-43 toxicity.
Loss of Brg1, BAF53b and CREST in spinal motor neurons of ALS patients
To establish relevance of findings in the ALS culture models, we examined autopsy cases of sporadic ALS with no identified mutation or cases with confirmed G4C2 expansion in C9ORF72 (Table 1); these forms of ALS are associated with WT TDP-43 (and occasionally FUS) sequestered in neuronal cytoplasmic inclusions (6–8). Indeed, nuclei of 100% of motor neurons in cross-sections of spinal cord from control cases were labelled by antibodies to Brg1, CREST and BAF53b, whereas these nBAF subunits were detected in only 42% of motor neurons from C9ORF72 cases and were depleted from 100% of motor neurons in samples from sALS cases (Fig. 6).
Additional information on autopsy cases examined for Figure 6
| Case | Diagnosis | ALS mutation | Sex | Age at death |
|---|---|---|---|---|
| 1 | Control (unknown non-neurological) | N/A | M | 67 |
| 2 | Control (Alzheimer's Disease/Diffuse Lewy Body Disease) | N/A | F | 91 |
| 3 | Control (heart disease) | N/A | M | 60 |
| 4 | Familial ALS | C9ORF72 | M | 58 |
| 5 | Familial ALS | C9ORF72 | M | 60 |
| 6 | Sporadic ALS | C9ORF72 | F | 49 |
| 7 | Sporadic ALS | C9ORF72 | F | 58 |
| 8 | Sporadic ALS | N/A | F | 71 |
| 9 | Sporadic ALS | N/A | F | 73 |
| 10 | Sporadic ALS | N/A | F | 63 |
| Case | Diagnosis | ALS mutation | Sex | Age at death |
|---|---|---|---|---|
| 1 | Control (unknown non-neurological) | N/A | M | 67 |
| 2 | Control (Alzheimer's Disease/Diffuse Lewy Body Disease) | N/A | F | 91 |
| 3 | Control (heart disease) | N/A | M | 60 |
| 4 | Familial ALS | C9ORF72 | M | 58 |
| 5 | Familial ALS | C9ORF72 | M | 60 |
| 6 | Sporadic ALS | C9ORF72 | F | 49 |
| 7 | Sporadic ALS | C9ORF72 | F | 58 |
| 8 | Sporadic ALS | N/A | F | 71 |
| 9 | Sporadic ALS | N/A | F | 73 |
| 10 | Sporadic ALS | N/A | F | 63 |
All autopsies were performed within 12 h post mortem.
Additional information on autopsy cases examined for Figure 6
| Case | Diagnosis | ALS mutation | Sex | Age at death |
|---|---|---|---|---|
| 1 | Control (unknown non-neurological) | N/A | M | 67 |
| 2 | Control (Alzheimer's Disease/Diffuse Lewy Body Disease) | N/A | F | 91 |
| 3 | Control (heart disease) | N/A | M | 60 |
| 4 | Familial ALS | C9ORF72 | M | 58 |
| 5 | Familial ALS | C9ORF72 | M | 60 |
| 6 | Sporadic ALS | C9ORF72 | F | 49 |
| 7 | Sporadic ALS | C9ORF72 | F | 58 |
| 8 | Sporadic ALS | N/A | F | 71 |
| 9 | Sporadic ALS | N/A | F | 73 |
| 10 | Sporadic ALS | N/A | F | 63 |
| Case | Diagnosis | ALS mutation | Sex | Age at death |
|---|---|---|---|---|
| 1 | Control (unknown non-neurological) | N/A | M | 67 |
| 2 | Control (Alzheimer's Disease/Diffuse Lewy Body Disease) | N/A | F | 91 |
| 3 | Control (heart disease) | N/A | M | 60 |
| 4 | Familial ALS | C9ORF72 | M | 58 |
| 5 | Familial ALS | C9ORF72 | M | 60 |
| 6 | Sporadic ALS | C9ORF72 | F | 49 |
| 7 | Sporadic ALS | C9ORF72 | F | 58 |
| 8 | Sporadic ALS | N/A | F | 71 |
| 9 | Sporadic ALS | N/A | F | 73 |
| 10 | Sporadic ALS | N/A | F | 63 |
All autopsies were performed within 12 h post mortem.
nBAF subunits are depleted from motor neurons in multiple forms of familial and sporadic ALS. Cross-sections of autopsied spinal cord from individuals dying of non-ALS causes (control), familial ALS owing to C9ORF72 G4C2 nucleotide expansion, or sporadic ALS with no known mutation were labelled with antibodies to the nBAF subunits (A) Brg1, (B) BAF53b or (C) CREST (red). Sections were labelled with SMI32 recognizing the high molecular weight neurofilament subunit (NFH) to identify motor neurons (green) and DAPI to label nuclei (blue). Arrow heads point to nuclei of motor neurons. Arrows point to accumulations of autofluorescent lipofuscin. Illustrated are representative images from examination of multiple sections of three cases per condition. Mean ± SD: 100 ± 0% (n= 3 control cases) and 12 ± 9% (n = 8 pooled C9ORF72 and sALS cases. Significant difference between control and ALS case groups, p = 0.0001). Scale bar = 16 µm.
Discussion
Brg1, BAF53b and CREST are required developmentally for neuronal differentiation and dendritic outgrowth (17,21,26,27) and mutations in nBAF subunits have been linked to autism spectrum disorders, which involve improper development and formation of dendrites (28–30). The present study demonstrated that Brg1 also is required for maintenance of mature dendrites as well as for neuronal development; inhibiting Brg1 function in cultured motor neurons with already developed dendrites (by treatment with the inhibitor PFI-3, expression of a dominant negative construct or knockdown of Brg1 gene expression) induced retraction of dendritic branches.
The loss of Brg1, BAF53b and CREST in ALS motor neurons identified in this study, along with the discovery of CREST mutations in ALS cases (20), implicates nBAF CRC in disease pathogenesis. Ectopic expression of Brg1 delayed mutant FUS or TDP-43-induced dendritic attrition in cultured motor neurons, linking nBAF loss to a common finding in ALS—attrition of dendritic architecture in upper and lower motor neurons (9–11).
Our previous study implicated transcriptional repression in FUS-induced motor neuron dysfunction (16): expression of mutant FUS in cultured motor neurons was associated with reduction in histone methylation and acetylation marks and reduction in total RNA synthesis, measured by BrU incorporation. However, the FISH data point to retention of Brg1 mRNA in the nucleus being a primary cause of the reduced expression at the protein level. Concomitant with loss of nuclear Brg1 protein, Brg1 mRNA was retained in the nucleus of cultured motor neurons with cytoplasmic FUS inclusions, implicating RNA processing or impaired transport of mRNA to the cytoplasm.
Previous studies have established that mutations in FUS disrupt its nucleocytoplasmic trafficking (22,31–33), and several effects on RNA metabolism have been described (34). Although the present study implicates inhibition of transport of Brg1 mRNA from the nucleus, the experimental model was not amenable to identifying whether this reflects disrupted interaction with FUS or some other mechanism, including nonspecific retention secondary to impaired nucleocytoplasmic transport. Previous studies have identified interaction of FUS with Brg1 (20), including our own unpublished data. PAR-CLIP seq data shows that FUS interacts with Brg1 (SMARCA4) RNA in intronic and 3' untranslated region (3’ UTR) regions for WT and mutant FUS, respectively (35). Recently, nuclear pore proteins involved in transport were identified as modifiers of the C9ORF72 phenotype (36–39) and cytoplasmic aggregates of disease-linked proteins (including TDP-43) were reported to interfere with nucleocytoplasmic transport of protein and RNA (40). The extent of involvement of various mRNAs and the similarity of mechanisms in multiple types of ALS remain to be resolved. However, the nuclear retention of Brg1 mRNA in motor neurons expressing mutant FUS highlights maintaining nucleocytoplasmic transport as an important therapeutic goal.
Consistent with our observations in ALS culture models, the critical nBAF CRC subunits Brg1, BAF53b and CREST were depleted from spinal motor neurons of ALS patients with C9ORF72 G4C2 expansions mutation and of sALS cases in which common ALS gene mutations had been excluded. Although the time course of nBAF changes cannot be surmised from autopsy specimens, motor neurons at different stages of disease are present in this tissue and the early changes in the culture models argues against disruption of nBAF CRC being only a late phenomenon in motor neuron degeneration. Regardless of the mechanisms underlying reduction in nBAF complexes, the downstream effects would have severe implications for neuronal function, including dysregulation of neuronal gene expression and loss of dendritic connections important for coordinating motor control. For motor neurons to function, they must retain their connections both centrally and peripherally. Evidence is accumulating that disruption of neuromuscular junctions is an early mechanism underlying loss of motor function (41); central connections are equally important, pointing to the potential use of epigenetic drugs targeting chromatin remodelling pathways in ALS.
Materials and Methods
Culture models of familial ALS
Primary culture models of FUS- and TDP-43-linked ALS were prepared using dissociated E13 CD1 mouse spinal cords as previously described (22,42). Briefly, spinal cords with attached dorsal root ganglia were excised from mouse embryos, dissociated in trypsin and plated on glass coverslips coated with poly-d-lysine and Matrigel® (Invitrogen Life Technologies, Burlington, ON, Canada) and cultured in modified N3, a hormone and growth factor enriched medium, supplemented with 2% horse serum (43). Cultures were used between 3 and 7 weeks after being established in order to allow for motor neuron maturation. Motor neurons are identified using phase microscopy by their large size (>20 µm) and large branching dendrites with bundles of neurofilaments; previous studies established that such neurons display characteristics of spinal motor neurons in vivo, including mature neurofilaments composed of the neurofilament triplet proteins without peripherin, expression of glutamate receptors and choline acetyltransferase, and firing of trains of action potentials (43).
To establish ALS models, plasmids encoding WT or mutant human FUS/TDP-43, in 50% Tris–EDTA buffer, were introduced into motor neurons by intranuclear microinjection at concentrations resulting in immuno-detectable protein expression in over 90% of neurons. For the microinjection procedure, cultures were bathed in Eagle’s Minimum Essential Medium (EMEM; Invitrogen Life Technologies) lacking bicarbonate, supplemented with 5 g/l glucose, and pH adjusted to 7.4. To identify injected neurons, mCherry plasmid or an inert 70 kDa dextran-fluorescein isothiocyanate marker was included in the injectate.
For each trial in culture models, injectates for each experimental group were prepared at the same time and microinjections for experimental groups were carried out consecutively, alternating one culture per group, beginning with a random sample. Cultures were tracked until the completion of the immunocytochemistry in order to maintain the appropriate treatments, but then were letter coded by an independent person for analysis.
Plasmids
Plasmids used in this study were: N-terminal Flag-tagged human FUS (WT or the mutants R521G, R521H and P525L) in pcDNA3 or eGFP-tagged constructs in eGFPN1 (22) injected at 20 ng/µl; human TARDBP (WT or the mutants G418C and A315T) with N-terminus Flag and C-terminus myc tags in pCS2+ at 20 ng/µl (42); pBJ5-Brg1 and pBJ5-Brg1 DN (dominant negative), a gift from Jerry Crabtree (Addgene #17873 and #17874) (25) at 20 ng/µl; SMARCA4 (encoding Brg1) and scramble shRNA expressing plasmids (pSuper-Brg1 and pSuper-scramble, kind gifts from Dr Betty Moran Rutgers University, NJ, USA) (24) at 30 ng/µl; mCherry-C1 (Clontech #632524) injected at 1–2 ng/µl. Note that all tagged proteins had previously been validated against untagged constructs.
Immunocytochemistry and immunohistochemistry
Cultures were fixed in 3% paraformaldehyde in phosphate buffered saline (PBS) for 10 min, permeabilized with 0.5% Nonidet-P40 in PBS and blocked in 5% horse serum in PBS. Cells were incubated in primary and secondary antibody solution for 60 and 30 min, respectively, with three 5-min washes in PBS after each incubation. Nuclei were counterstained with Hoechst. Images were acquired using a Zeiss Observer Z1 microscope (Carl Zeiss Canada Ltd, Toronto, ON, Canada) equipped with a Hamamatsu ORCA-ER cooled CD camera (Hamamatsu, Japan) and filter sets 02, 13 and 00 (Carl Zeiss Canada Ltd).
Levels of nBAF subunits in the nucleus were measured semi-quantitatively as intensity of secondary antibody epifluorescence. Images were acquired at equal exposures and fluorescence intensity profiles were analysed on the raw images with Axiovision software (Carl Zeiss Canada Ltd) by tracing the region of the nucleus with the outline spline function. In cases where image contrast/brightness was altered to improve visibility in figures, the same parameters were applied to all figure components.
Human autopsy material (ALS and control cases) was obtained from the tissue repository at Sunnybrook Health Sciences Centre, Toronto, ON, Canada. ALS cases were diagnosed using the Revised El Escorial Criteria, and patient consents were obtained from the legal representatives in accordance with the Ethical Review Boards of Sunnybrook Health Sciences Centre and University of Toronto (Toronto, Canada). Control and ALS cases were processed according to the same standardized procedures and were genotyped for common ALS genes [TARDBP, FUS, superoxide dismutase 1 (SOD1), C9ORF72]. Further details of cases examined are presented in Table 1. Unfortunately, familial ALS cases with FUS or TARDBP mutations were not available in this series.
For immunohistochemistry, 4 μM serial sections of formalin-fixed, paraffin-embedded ALS and control lumbar spinal cords were mounted on glass slides then deparaffinized and rehydrated with xylene and graded ethanol washes, and final wash in water. For antigen retrieval, slides were incubated for 10 min at 95° C in 10 mM sodium citrate, 0.05% Tween 20, pH 6 then blocked for 2 h at room temperature in 10% goat serum, 3% bovine serum albumin and 0.3% Triton X-100 in Tris-buffered saline (TBS). Primary antibody in antibody diluent (DAKO, S3622) was applied overnight at 4°C, followed by three 5-min washes in TBS and 0.05% Tween 20 (TBST) followed by secondary antibody for 40 min at room temperature, then three 5-min washes in TBST and mounting in Prolong Gold antifade reagent with DAPI (Invitrogen, P36931). Epifluorescence images were obtained using a Leica DMI6000B inverted microscope (Leica Microsystems Inc., Concord, ON, Canada) and Volocity 6.3 imaging software (PerkinElmer, Waltham, MA, USA). As postmortem changes and fixation can affect the ability to detect proteins in autopsy specimens by immunohistochemistry, antibodies were pre-tested in mouse tissue under mimicked postmortem conditions, as previously described (16).
Antibodies
Antibodies used were: Mouse anti-FLAG M2 (Sigma-Aldrich, St. Louis, MO, USA, F1804, 1:400); mouse anti-FUS (Santa-Cruz Biotechnology, Inc., Dallas, TX, USA sc-47711, 1:100); rabbit anti-FUS (Proteintech, Chicago, IL, USA, 11570-1-AP, 1:400); rabbit anti-Brg1 (Proteintech, Chicago, IL, USA, 21634-1-AP, 1:600); rabbit anti-BAF53b (Abcam ab140642, 1:400); rabbit anti-CREST (Proteintech 12439-1-AP, 1:800); rabbit anti-TDP-43 (ProteinTech 10782-2-AP, 1:500); mouse anti-200 kDa neurofilament (Abcam SMI32, 1:500); goat anti-digoxigenin (Abcam ab76907, 1/300); anti-mouse Cy2-IgG (Rockland, Gilbertsville, PA, USA, 610-711-124, 1:300); anti-rabbit IgG-Cy3 (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA, 715-165-150, 1:300), anti-rabbit IgG-Cy2 or -Cy3 (Jackson ImmunoResearch Laboratories, Inc., 711-225-152; 711-165-152, 1:300), anti-rabbit IgG-Alexa Fluor 188, anti-mouse IgG-Alexa Fluor 594 (Life Technologies), and anti-goat IgG-Cy3, 715-185-147, 1:300 (Jackson ImmunoResearch Laboratories).
Analysis of dendritic morphology
For Sholl analysis (44,45), motor neurons were microinjected with mCherry plasmid vector to visualize entire cell morphology. Images of mCherry epifluorescence were taken at 100×, at exposures revealing the extent of dendritic branches, using a Zeiss Observer Z1 microscope (Carl Zeiss Canada Ltd) equipped with a Hamamatsu ORCA-ER cooled CD camera (Hamamatsu, Japan) and filter sets 02, 13 and 00 (Carl Zeiss Canada Ltd).
Images were exported to 8-bit .tif images and the cell body and dendrites were traced using the semi-automated ImageJ tracing plugin, NeuronJ (http://www.imagescience.org/meijering/software/neuronj/). Branching points were designated using NeuronStudio Documentation (Mount Sinai School of Medicine) and measurements of dendritic morphology were performed with Bonfire (44) using the resulting .swc file. In order to control for bias, evaluators were blinded to conditions for image tracing.
Non-radioactive FISH
Sense (control) and anti-sense riboprobes to reveal Brg1 mRNA were generated as previously described (46) from cDNA obtained by reverse-transcription of mRNA extracted from the cerebral cortex of an adult CD1 mouse, using the following primers: Forward—GTAGCTCAGTGGATGCATGC, Reverse—TGAAGCAGGGTCTTGTCACT. Non-radioactive FISH combined with immunocytochemistry was carried out according to Tiveron et al. (47) with some modifications. Probes were labelled with digoxigenin-11-UTP. Spinal cord cultures were fixed in 4% paraformaldehyde in PBS for 30 min, then permeabilized in 0.3% Triton X-100 in 4% PFA for 5 min. Cover slips were washed two times in PBS followed by one wash in RIPA buffer for 5 min. Residual aldehyde moieties were quenched as described in Tiveron et al. (47), using triethanolamine buffer with 0.25% acetic anhydride for 15 min followed by washes in PBS. Pre-hybridization was carried out in hybridization solution (formamide 50% final, SSC 5×, Denhardt’s solution 5×, Salmon Sperm DNA 500 µg/ml, yeast RNA 250 µg/ml). The probe was then diluted to 1 µg/200 µl and incubated overnight at 70°C followed by two washes of 30 min at 70°C in freshly prepared, pre-warmed post-hybridization solution (formamide 50% final, SSC 2×, 0.1% Tween-20, followed by incubation in MAB buffer (100 mM maleic acid, 150 mM NaCl). Flag-tagged FUS and digoxigenin-tagged probes were detected by immunocytochemistry with anti-Flag and anti-digoxigenin antibodies (see above) diluted in MAB buffer containing 5% horse serum. Images were recorded as above, with any contrast enhancement being applied to all figure components.
Statistical analysis
Mean ± S.E.M. were calculated and differences between groups were analysed as identified in the figure legends, with significance established at P < 0.05. Each experiment was conducted using a minimum of three cultures per condition from the same culture batch, and subsequently was repeated in two additional culture batches.
Supplementary Material
Supplementary Material is available at HMG online.
Conflict of Interest statement. None declared.
Funding
This work was supported by research grants from the Amyotrophic Lateral Sclerosis Association to H.D. and J.R.; the Muscular Dystrophy Association to H.D.; the Canadian Institutes for Health Research, the ALS Society of Canada and les Fonds de recherche du Québec-Santé under the frame of E-Rare-2, the ERA-Net for Research on Rare Diseases to H.D.; the Canadian Institutes for Health Research to J.R.; the W. Garfield Weston Foundation to E.R. and L.Z.; a Starting Grant from the European Research Council to C.R.; a fellowship from Inserm & Région Alsace to C.M. and a Milton Safenowitz Fellowship from the Amyotrophic Lateral Sclerosis Association to B.Z.
