Background
Mutations in the gene encoding fused in sarcoma/translocated in liposarcoma (FUS/TLS or FUS), also known as the heterogeneous nuclear ribonucleoprotein (hnRNP) P2 [
1], are linked to inherited cases of amyotrophic lateral sclerosis (ALS) [
2,
3]. ALS is a fatal neurodegenerative disease characterized by motor neuron loss, progressive muscle weakening and paralysis [
4]. Most ALS-linked FUS mutations are located within the C-terminal nuclear localization signal (NLS) that binds transportin, the nuclear importer that translocates FUS from the cytoplasm into the nucleus [
5,
6]. Although FUS is predominately localized to the nucleus in most cell types [
7], it has nucleo-cytoplasmic shuttling capabilities that may be important for mRNA transport [
8]. In fact, FUS is thought to play a role in local translation at the dendrites of neuronal cells [
9‐
11]. Disruption of the FUS/transportin interaction leads to nuclear depletion with concomitant cytoplasmic accumulation of FUS in cultured mammalian cells [
6]. The potential relevance of this interaction is underscored by the cytoplasmic accumulation of FUS in both ALS [
2,
3] and frontotemporal lobar degeneration (FTLD) [
12] post-mortem central nervous system (CNS) tissues.
The extent to which FUS mutants redistribute to the cytoplasm correlates with ALS disease severity [
6,
13]. For example, individuals with the FUS R495X mutation, which leads to truncation of the NLS and significant cytoplasmic retention of FUS, exhibit early disease onset and a relatively severe disease course [
13]. Nuclear depletion of FUS may impair putative nuclear functions involving mRNA [
14,
15] and DNA [
16,
17] processing. An alternative, though not mutually exclusive, possibility is that mutant-FUS exerts a gain-of-toxic function in the cytoplasm [
18].
Recently, a two-hit model has been proposed to account for cytoplasmic FUS toxicity in ALS and FTLD [
19]. Cytoplasmic mislocalization of FUS, either through genetic mutations or other unidentified factors, represents the first hit. The first hit alone may not be sufficient to cause disease. However, a second hit stemming from cellular stress directs cytoplasmic FUS into stress granules. Stress granules are stalled translational complexes that form as a normal response to induced stressors such as oxidation, heat-shock, viral infection or hypoxia [
20]. The function of stress granules is thought to be in the triage of mRNAs that are destined for expression, storage or degradation, which in turn restores cellular homeostasis [
21]. Stress granule function may not be limited to mRNA processing, as the activity of certain proteins can also be controlled by their sequestration into stress granules [
22]. It follows that the association of mutant-FUS with stress granules may impair stress response and ultimately cause disease [
23]. This notion is supported by evidence of stress granule marker proteins within the pathological aggregates of neurodegenerative disease tissues [
6,
24,
25].
While we and others have firmly established a mutant-specific phenotype with respect to FUS in stress granules [
5,
6,
13,
26‐
30], there is little evidence that mutant-FUS actually alters the properties of stress granules. Although there is no functional assay per se for stress granules, the properties of stress granules that are thought to be relevant to their function include assembly kinetics, dynamics, morphology and abundance [
21,
31]. Under conditions of oxidative stress, we show that mutant-FUS delays stress granule formation in mammalian cell culture. Once sodium arsenite-induced stress granules are formed, however, those containing mutant-FUS are more dynamic, larger and more abundant compared to stress granules lacking FUS. Upon removal of stress, stress granules disassemble more rapidly in cells expressing cytoplasmic mutant-FUS. Further, we identified the RGG domains within FUS as playing a key role in the assembly of mutant-FUS into stress granules, although the methylation of arginine residues within these RGG domains does not play a significant role. The evidence presented here supports the hypothesis that the association of mutant-FUS with stress granules represents a gain-of-toxic interaction in ALS pathogenesis.
Discussion
The association of cytoplasmically mislocalized ALS-linked FUS mutants with stress granules is well established [
5,
6,
13,
26‐
30], but what affect does mutant-FUS have on the properties of stress granules? We sought to examine the effect of mutant-FUS on the physical properties of stress granules that are potentially linked to function. While there is no functional assay per se for stress granules, they are believed to represent sites of mRNA triage, which influences whether particular mRNA transcripts are retained within stress granules, translated on ribosomes or degraded within P-bodies [
20]. There is also evidence that the signaling activity of proteins can be controlled by their sequestration and/or release from stress granules during stress [
22]. Thus, the cellular response to stress is modulated, at least in part, by stress granules. Since mutant-FUS, but not WT-FUS, is incorporated into stress granules under various induced stressors, the mutant protein has the potential to disturb stress granules and impair cellular stress response in ways that could contribute to ALS pathogenesis [
19].
Stress granules assemble in response to induced stress. Our results show that ALS-linked, cytoplasmic FUS R495X delays the assembly of stress granules in both HEK-293 (Figure
1) and neuronal NSC-34 (Figure
2) cells under conditions of acute oxidative stress. The predominantly nuclear FUS H517Q mutant also delays stress granule assembly, but to a lesser degree than FUS R495X (Figure
1). Therefore, the delay in stress granule assembly correlates with cytoplasmic levels of mutant-FUS, probably because the protein is poised to enter stress granules once stress is induced. Since over-expression of some ALS-FUS mutants reportedly induce the spontaneous formation of cytoplasmic inclusions that stain positively for stress granule markers [
28,
29], one might expect the expression of mutant-FUS to correlate with a faster rate of stress granule assembly. However, the properties of stress granules are influenced by the nature of the induced stressor [
48], and we show that stress granules induced by protein over-expression exhibit different dynamic properties than those induced by sodium arsenite (Additional file
3). We also demonstrate that mutant-FUS accelerates the disassembly of stress granules (Figure
1D). Therefore, expression of mutant-FUS appears to disfavor the formation of and/or destabilizes stress granules, possibly by interfering with protein interactions within these structures (Figure
3). The effects of mutant-FUS on stress granule assembly and disassembly are reminiscent of effects seen during TDP-43 knock-down [
37]. Considering that stress granule assembly is a regulated process [
20], factors that either delay or accelerate stress granule assembly/disassembly may adversely affect cellular homeostasis.
Interestingly, once stress granules are formed, mutant-FUS exerts an effect on both stress granule morphology and abundance that may appear counterintuitive based on the effect of FUS during the processes of assembly and disassembly. While the expression of GFP-FUS R495X both disfavors stress granule assembly and weakens stress granule associated interactions, under conditions of persistent stress the size and abundance of stress granules is augmented by the expression of mutant-FUS (Figure
4). This increased size and abundance of stress granules does not necessarily mean these structures are held together more tightly, but rather is a likely consequence of the additional protein load associated with these structures from the GFP-FUS R495X protein itself. This rationale may also be relevant to the increased size of stress granules in ALS-linked TDP-43 mutants under conditions of hyperosmolar stress [
49] and suggests that this phenotype may be part of a common disease pathway. An intriguing, but not mutually exclusive, possibility is that mutant-FUS and TDP-43 recruit additional protein partners and mRNA substrates into stress granules, thereby further increasing their size and abundance (Figure
4 and [
25,
30]). Indeed, thousands of mRNA transcripts are bound by FUS [
58,
59] with many distinct mRNAs bound by cytoplasmic mutant-FUS but not WT FUS [
59]. Therefore, mutant-FUS may inappropriately process mRNAs and/or facilitate aberrant cytoplasmic protein interactions during stress. The latter possibility is supported by our FRAP analyses, which showed that both mRFP-G3BP and mRFP-TIA-1 exhibit weaker binding and heightened dynamics within sodium arsenite-induced stress granules containing mutant-FUS (Figure
3). In fact, GFP-FUS R495X altered the dynamic properties of stress granules in all of our FRAP experiments, raising the possibility that mutant-FUS interferes with the sorting mechanisms [
21,
22] associated with these structures under stress.
If the association of mutant-FUS with stress granules does indeed represent a gain of toxic interaction, it will be important to identify factors that modulate this association. Although FUS contains multiple domains that contribute to FUS aggregation [
60] and/or are homologous to sequences that direct other proteins into stress granules [
26,
50‐
52], our results show that the RGG domains are largely responsible for directing FUS into stress granules (Figure
5). Our results are in general agreement with a recent report by Bentmann et al., which also demonstrated a key role for the RGG domains in assembling FUS into stress granules [
26]. However, our results do not support a role for the Gly-rich and RRM domains in this process, whereas the former study did. This discrepancy may be due to the difference in FUS constructs, stressor (sodium arsenite versus heat shock [
26]) and/or the FUS mutation (R521G versus P525L [
26]) that were employed in these studies. Whether the RNA-binding ability of FUS is required for its localization to stress granules is not altogether clear [
26,
27]. Several domains within FUS exhibit RNA-binding capabilities, including the RMM, RGG, and zinc finger domains. Bentmann et al. demonstrated a correlation between cytoplasmic FUS constructs that bound RNA and were incorporated into stress granules, consistent with a role for RNA-binding in the assembly of mutant-FUS into stress granules [
26].
That the RGG domains direct mutant-FUS to stress granules raises the possibly that this process is controlled by arginine dimethylation of RGG motifs [
61]. Emerging evidence indicates that the RGG motifs within FUS are methylated by protein arginine N-methyltransferase-1 (PRMT1) [
55,
62,
63], and that this post-translational modification can influence the subcellular localization of mutant-FUS [
62,
63]. While stress granules contain methylated proteins (Figure
6B), and methylated forms of mutant-FUS have been detected in both stress granules and diseased-tissues [
5], our data suggests that methylation of FUS is not a prerequisite for its incorporation into stress granules (Figure
6).
How might the incorporation of mutant-FUS into stress granules alter cellular homeostasis under conditions of induced stress, and what are the implications for neurodegenerative disease? We show that mutant-FUS delays stress granule assembly (Figures
1 and
2), decreases the binding of stress granule-associated proteins within stress granules (Figure
3), and increases both size and abundance of stress granules (Figure
4). These physical and dynamic properties of stress granules are thought to be linked to stress granule function, and thus the effects of mutant-FUS may culminate in impaired stress response and, eventually, in neurodegeneration. Although there have been no reports of overt cytotoxicity in mutant-FUS cellular models exposed to sodium arsenite or other stressors, the effects of impaired stress response may appear more distinctly as a function of age, disease progression and/or chronic stress in the human disease [
19]. In fact, stress granule marker proteins have been detected within the pathological inclusions of CNS tissues from patients with ALS and FTLD [
6,
25], supporting the notion that stress response factors are altered during the course of disease. Moreover, these observations raise the possibility that stress granules are precursors to the end-stage aggregates that are characteristic of these diseases [
23]. Although our data show that mutant-FUS accelerates stress granule disassembly, under conditions of persistent stress these granules containing mutant-FUS are larger and more numerous and thus have the potential to coalesce into larger aggregates. Extending analyses of stress granules to other models systems, such as human iPS cells from individuals with ALS or ALS rodent models, may allow us to better address whether altered stress granule assembly plays a role in disease onset and/or progression, and whether the association of ALS-linked proteins with stress granules does in fact impact disease.
Materials and methods
Cell culture and drug treatments
Inducible GFP-FUS expressing FlpIn HEK-293 cells were maintained as described previously [
13]. Human cervical carcinoma cells (HeLa) were maintained in Modified Eagle’s medium (MEM, Gibco 10370) supplemented with 10% (v/v) heat inactivated fetal bovine serum (Sigma, F4135), 2 mM L-glutamine (Gibco, 25030), and 1% (v/v) penicillin and streptomycin solution (Gibco, 15140). Mouse motor neuron–like hybrid cell lines (NSC-34) [
38] constitutively expressing untagged human FUS were maintained in Dulbecco’s modified Eagle’s medium (Invitrogen, 11965118) supplemented with 10% (v/v) tetracycline-tested fetal bovine serum (Sigma, F6178), and 2 μg/mL puromycin (Invitrogen, A11138-03).
NSC-34 cells constitutively expressing untagged human FUS constructs were generated by lentiviral transduction of the CSCW2-IRES-GFP lentivector (a generous gift from Dr. Miguel Esteves, University of Massachusetts Medical School) containing FUS (WT or mutant R495X). Flow cytometry was used to enrich for expression of the GFP reporter in each line. Cells with equivalent levels of exogenous FUS proteins were employed.
For drug treatments, the following stocks were prepared and stored at freezing temperatures: 50 mg/mL doxycycline (Sigma, D9891) in water (−80°C), 100 mM sodium arsenite (Sigma, 71287) in water (−20°C) and 20 mM adenosine-2,3 dialdehyde (“AdOx”; Sigma, A7154) in water (−20°C). FUS expression in the FlpIn HEK-293 lines was induced with the addition of 1 μg/mL doxycycline for 24 hrs unless otherwise noted. Cells were then exposed to sodium arsenite and/or AdOx as described below.
Plasmids and cloning
The pEmRFP-G3BP and mRFP-TIA-1 plasmids for FRAP analyses were generously provided by Drs. Nancy Kedersha and Paul Anderson (Brigham and Women’s Hospital, Harvard Medical School). The mRFP-TIA-1 was sub-cloned into the low expression lentivirus vector CShPW2 (a gift from Miguel Estevez, UMMS) with NheI and Knp1 restriction sites using BP Clonase II (Invitrogen, 11789–020), thereby creating CShPW2-RFP-TIA-1. MBP-M9M was a kind gift from Dr. Yuh Min Chook (University of Texas Southwestern Medical Center).
GFP-FUS R521G deletion constructs were constructed as follows: PCR amplified full length GFP-FUS R521G, flanked by attB homologous sequences, was cloned into pDONR221 vector (Invitrogen, 12536–017) with BP Clonase II (Invitrogen, 11789–020) to generate the starting plasmid pDONR221:GFP-FUS R521G. To facilitate substitution of the full length gene with deletion/truncation variants, restriction sites for KpnI and XbaI were introduced upstream of the ATG start codon and downstream the TAA stop codon, respectively, using the following primers: fwd: GGGGACAAGTTTGTACAAAAAAGCAGGCTGGTACCATGGCCTCAAACGATTATACCC, rev: GGGGACCACTTTGTACAAGAAAGCTGGGTTCTAGATTAATACGGCCTCTCCCTGC. To generate deletion constructs, the following primers were designed by joining the upstream and downstream sequences flanking the domain that was deleted: ∆GLY_fwd: AGAACCAGTACAACAGCAGCAGTACCATCTTTGTGCAAGGCC; ∆GLY_rev: ACTCAATTGTAACATTCTCACCCAGACTGCCAGACAACAACACCCGGGCAGACTTTAATCGGG; ∆RRM_rev: CCACGACCATTGCCACCACCGTTGTTGTCTGAATTATCCTGTTCG; ∆RGG1_fwd: CAAGGTCTCATTTGCTACTCGCGCTGGTGACTGGAAGTGTCC; ∆RGG1_rev: CATATTCTCACAGGTGGGATTAGGCCGATTAAAGTCTGCCCGGC; ∆RGG2_fwd: CCAGTGTAAGGCCCCTAAACCAGATAAGATGGATTCCAGGGGTGAGCAC; ∆RGG2_rev: GTGCTCACCCCTGGAATCCATCTTATCTGGTTTAGGGGCCTTACACTGG; ∆422-526_fwd: GGGGACAAGTTTGTACAAAAAAGCAGGCTGGTACCATGGCCTCAAACGATTATACCC; ∆422-526_rev: GGGGACCACTTTGTACAAGAAAGCTGGGTTCTAGATTATCGCTGCTGTCCTCCACC. The deletion reactions were performed using the pDONR221:FUS R521G plasmid as template and the QuikChange II Mutagenesis kit (Stratagene; 200523) according to the manufacturer’s instructions. For the ∆QGSY truncation construct, PCR was performed using a reverse primer for the full-length R521G gene paired with a forward primer containing the 5’-end sequences of ∆GLY flanked by the restriction enzyme KpnI recognition sequence: ∆QGSY_fwd: CAGGCTGGTACCGGTGGTGGAGGTGGAGGT. All constructs were then sub-cloned into the expression vector pDEST-53 (Invitrogen) using Gateway cloning method with LR Clonase II (Invitrogen, 11791–100) according to the manufacturer’s instructions.
Immunofluorescence
Standard immunofluorescence protocols were employed as described previously [
13]. Briefly, cells were fixed with 4% paraformaldehyde for 10–15 min then blocked with PBSAT (1X PBS/1% BSA/0.5% Triton-X 100) for 30–60 min at ambient temperature. Primary antibodies described in each experiment were diluted in PBSAT and applied to cells at ambient temperature for 1 hr. Primary antibody dilutions were as follows: 1:2000 for mouse anti-G3BP (BD Transduction Labs, 611126), 1:1000 for rabbit anti-G3BP (Proteintech, 130-57-2AP), 1:1500 for rabbit anti-dimethyl arginine (“ASYM24”; Millipore, 07–414) and 1:200 mouse anti-FUS (Santa Cruz, SC-4771). Cells were then incubated with secondary antibodies diluted 1:1000–1:2500 in PBSAT for 45 min at ambient temperature. Secondary antibodies included Dylight 549 conjugated anti-mouse IgG (Jackson ImmunoResearch Labs, 715-505-151), Cy3 conjugated anti-mouse IgG (Jackson ImmunoResearch Labs, 715-165-151), Cy3 conjugated anti-rabbit IgG (Jackson ImmunoResearch Labs, 711-165-152), and Cy5 conjugated anti-mouse IgG (Jackson ImmunoResearch Labs, 715-175-151). GFP signal was enhanced by 1:1000 dilution of Alexa Fluor 488-conjugated rabbit anti-GFP (Invitrogen, A21311). Cells were stained with 34 ng/mL DAPI in dH
2O, and coverslips were mounted with ProLong Gold anti-fade reagent (Invitrogen, P36930).
Western blotting
Standard western blotting protocols were employed as described previously [
13]. Primary antibodies described in each experiment were diluted as follows: 1:1000 for mouse anti-GFP (Living Colors; Clontech, 632380), 1:1000 for rabbit anti-dimethyl-arginine (“ASYM24”; Millipore, 07–414), 1:1000 for mouse anti-tubulin (Sigma, T9026), and 1:1000 for rabbit anti-FUS. Rabbit anti-FUS antibodies were generated by GenScript against a C-terminal epitope, using the peptide CKFGGPRDQGSRHDSEQDNSD. Blots were incubated with primary antibodies for 1 hr at ambient temperature or overnight at 4°C. Secondary antibodies, including anti-mouse IRDye 680 (Licor, 926–32220) or IRDye 800 (LiCor, 926–32210) and anti-rabbit IRDye 680 (LiCor, 926–32220) or IRDye 800 (Licor, 926–32211), were diluted 1:10000 and incubated with blots for 1–2 hrs at ambient temperature. Bands were visualized with an Odyssey Infrared Imager (LiCor, Model 9120), and densitometry measurements performed with the Odyssey Software (LiCor, V3.0).
Stress granule assembly and disassembly kinetics
Inducible GFP-FUS (WT, R495X and H517Q) HEK-293 cells were plated on coverslips at a density of 5 × 104 cells/coverslip. The next day, GFP-FUS expression was induced as described above. For stress granule assembly measurements, cells were treated with 0.25 mM sodium arsenite for 40 or 90 min. Coverslips were then fixed and processed for immunofluorescence (IF) with the mouse anti-G3BP and rabbit anti-GFP antibodies listed above. For stress granule disassembly measurements, cells were treated with 0.25 mM sodium arsenite for 60 min, at which time the media containing sodium arsenite was replaced with fresh media. After 90 min in fresh media, cells were processed for IF as described above. The percentage of cells with stress granules was determined as [(the number of cells containing at least one stress granule / total number of cells) × 100]. More than 2000 healthy, interphase, non-crowded cells were counted in multiple (between n=4 and n=11, depending on cell line and condition) independent experiments for both assembly and disassembly conditions. A one-way ANOVA with Tukey's multiple comparisons post-test was used to compare the induced and uninduced groups. Similar parameters were used for assembly kinetics in NSC-34 parent cells and cells expressing untagged human FUS proteins with the following changes: cells were treated with sodium arsenite for 1 hr, and immunofluorescence was performed with the rabbit anti-G3BP and mouse anti-FUS antibodies listed above. More than 2000 cells were counted in at least n=11 independent experiments per condition. Statistical significance was determined by one-way ANOVA with Tukey's multiple comparisons post-test.
Fluorescence recovery after photobleaching
GFP-FUS WT and GFP-FUS R495X HEK-293 cells were plated at a density of 8 ×104 cells/plate in 35 mm glass bottom dishes (MatTek Corp, P35GC-1.5-14-C), allowed to adhere for 48 hrs, then transfected with either CShPW2-RFP-TIA-1 or pEmRFP-G3BP expression plasmids using Lipofectamine 2000 (Invitrogen, 11668–019) according to the manufacturer’s instructions with a 1.6 μl Lipo: 3.2 μg DNA ratio. Approximately 23 hr post-transfection, phenol red(−) growth media without (overexpression experiments) or containing 0.2 mM sodium arsenite (stress granule experiments) was applied to the cells for 1 hr prior to FRAP.
FRAP experiments were performed at 37°C as previously described [
45]. Multiple cells were analyzed in each experiment over a 30 min period starting at 60 min of sodium arsenite exposure. Experiments were carried out on a Leica SP5 AOBS laser scanning confocal microscope using a 40× 1.3NA water immersion objective or a Leica SP1 system using a 40×, 1.25NA oil immersion objective. No more than two stress granules, from opposite sides of a cell, were individually bleached using a 1-3s laser pulse delivered by a 488 nm or 561 nm laser. A pre-bleach, immediate post-bleach and 16 additional post-bleach images spaced at 5 sec intervals were captured. Leica Confocal Software (Leica Microsystems, Exton, PA) was used to measure fluorescence intensity in the bleached region of interest (ROI), the whole cell, and in a background control area lacking cells at each time point. The data was analyzed and background fluorescence subtracted using Excel. The relative fluorescence intensity (I
rel) in the bleached area was calculated as previously shown [
45]. Briefly, the following equation was used: I
rel, t = (I
t × (C
0/C
t))-(I
pbl × (C
0/C
bl))/(I
0-(I
pbl × (C
0/C
pbl)), where C
0 is the total cellular fluorescence before bleaching, C
pbl is the total cellular fluorescence in the post-bleach image, C
t is the total cellular fluorescence at time t, I
0 is the pre-bleach ROI fluorescence intensity, I
t is the ROI fluorescence intensity at time t, and I
pbl is the post-bleach ROI fluorescence intensity. The data was normalized using this equation such that the post-bleach ROI fluorescence intensity was set to 0 and the pre-bleach ROI fluorescence intensity to 1. I
t was calculated as the percentage difference between the relative fluorescence asymptote of the recovery curve and a relative recovery of 1, a value that would reflect complete recovery without an immobile fraction. Recovery curves were drawn using Graphpad Prism 6 (Graphpad Software), with individual time points presented as means ± SEMs. Fluorescence recovery half times were calculated from exponential one-phase association curves best fit for the recovery graphs: F(t) = F
max (1- e
-kt). Mobile fractions were calculated from the plateau region from each curve, which was identified as the series of data points with < 2% change in slope over time.
At least two independent experiments were performed for all conditions. The total number of stress granules analyzed for each condition is shown in Figure
3. A two-way ANOVA was used to determine statistical significance between the mobile fractions for each of the experiments.
Morphology experiments
Stable GFP-FUS HEK-293 cells were plated and processed for immunofluorescence as described under ‘stress granule assembly and disassembly kinetics’ above, except that cells were treated with 0.5 mM sodium arsenite for 1 hr. Confocal stacked images (0.2 μm stack step, 4 μm range) were acquired using a Zeiss Axiovert 200 microscope with a PerkinElmer UltraView LAS spinning disc equipped with a 100× phase objective. Imaris analytical software (Bitplane Scientific Software) was used to construct 3D projections of image stacks. Volume measurements were taken of each stress granule with a G3BP fluorescence signal that was at least 2-fold above background. Because P-bodies had an average volume of 0.5 μm3 (data not shown) by the same analysis, only stress granules with a volume > 0.5 μm3 were included in the analysis. Data is averaged from three independent experiments per line, using approximately 30 stress granules per condition. Statistical significance was determined using the Student’s t-test.
Analysis of FUS deletion constructs in stress granule assembly
HeLa cells were plated on coverslips at a density of 2.5 × 104 cells/coverslip and adhered at 37˚C for 24 hrs, after which GFP-FUS R521G truncation constructs and MBP-M9M were transiently transfected into cells using Lipofectamine-2000 (Invitrogen, 11668) according to the manufacturer’s instructions. Twenty-four hours post transfection, cells were subjected to media containing 0.5 mM sodium arsenite for 1 hr. Cells were processed immunofluorescence with mouse anti-G3BP as described above.
Confocal microscopy was performed using a Solamere Technology Group CSU10B spinning disk confocal system equipped with a Yokogawa CSU10 spinning disk confocal scan head. Image stacks (0.2 μm stack step; 13 stack range) were acquired using a 100× oil objective, a Roper Cool-snap HQ2 camera and MetaMorph V7.6.3 software. Background signal was subtracted by removing fluorescence from a dark-current image (acquired with the laser off) from each raw image. For the GFP images, variations in illumination and detection efficiencies at each pixel were corrected by dividing the dark-adjusted intensities by a normalized flat-field image of a uniformly green fluorescent slide (Chroma Technology, Rockingham, VT, USA) acquired using the same 525/50 nm band-pass filter. Four channels were imaged per cell: FITC for GFP-FUS, Cy3 for G3BP, DAPI for nuclei and phase for cell borders.
The extent of GFP-FUS incorporation into stress granules was analyzed with MetaMorph V7.6.3 software using the Integrated Morphometry Analysis tool. Since cells with a GFP signal brighter than 1.5 × 106 grays/μm2 tended to form cytoplasmic aggregates a priori of arsenite treatment (data not shown), only transfected cells with GFP-FUS expression levels < 1.5 × 106 grays/μm2 were selected for analysis. Stress granules were selected using the Cy3 (G3BP) channel as a reference. The slice corresponding to the center of the stress granule was selected from each image stack. Stress granules with an area of at least 0.5 μm2 were selected. An outline was drawn around each G3BP granule, and this outline was then transferred to the corresponding FITC (GFP-FUS) image, such that the GFP-FUS signal intensity (“stress granules intensity”) within that granule could be measured. GFP signal intensity measurements were also acquired in the region proximal to the stress granule, and was referred to as the “Diffuse intensity”. The ratio of stress granule intensity (i.e., GFP-FUS inside the stress granule) to diffuse intensity (i.e., GFP-FUS outside the stress granule) was determined for a total of 75–150 stress granules per construct over three independent experiments. Statistical significance was determined by a one-way ANOVA followed by a Dunnett’s post hoc test in Graphpad Prism 6 (Graphpad software).
Methyltransferase inhibition studies
HeLa cells were plated at a density of 2.5 × 104 cells/coverslip and adhered for 24 hrs, after which GFP-FUS R495X expression constructs were transiently transfected with Lipofectamine-2000 either with or without 25 μM AdOx for 24 hr. Cells were then exposed to 0.5 mM sodium arsenite for 1 hr, and coverslips were processed for immunofluorescence using the mouse anti-G3BP rabbit and ASYM24 antibodies as described above. For quantification, confocal images of 30 cells per condition across three independent experiments were taken, and the intensity of GFP-FUS and ASYM24 signal within stress granules was determined between AdOx-treated and untreated conditions using MetaMorph as described above. Statistical significance was determined using a Student’s t-test.
For GFP immunoprecipitation (IP) reactions, cells were lysed for 15 min in IP buffer (400 μL 1% NP-40 (MP Biomedicals, 198596)/50 mM Tris–HCl (Sigma T3253-500G)/5 mM EDTA (Fisher E478-500)/150 mM NaCl and 10% v/v glycerol (Acros 15982–0010) in water; pH 7.5), and centrifuged at 13000 rpm for 15 min at 4°C. The supernatant was pre-cleared with 50 μL Biomag Protein G beads (Qiagen, 311812) for 2 hrs at 4°C. Anti-GFP-coated beads for each sample were prepared by incubating 0.5 μL of anti-GFP (Abcam, ab290) in 400 μL of IP buffer with 50 μL Biomag Protein G beads for 2 hrs at 4°C. IP reactions were performed at 4°C overnight with 100 μg of pre-cleared lysate. Protein elution was accomplished with 50 μL 1X SDS loading buffer (Boston Bioproducts BP11R) for 5 min at 95°C, and 20 μL of sample was subjected to western blot analysis with mouse anti-GFP, mouse anti-tubulin and ASYM24 as described above.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
DB, LK, and CW planned and performed the majority of experiments; DB, AJQ and JAN planned, performed and analyzed data for FRAP; RJC cloned deletion constructs for structure-function analyses; RRKS and KB contributed to the design and data interpretation for experiments; DB, LK and DAB wrote the paper. All authors read and approved the final manuscript.