Drosophila Ref1/ALYREF regulates transcription and toxicity associated with ALS/FTD disease etiologies

We established an in vivo RNAi screen in Drosophila melanogaster to identify RNA interacting proteins that can modify the toxicity of ALS-associated genes (Fig. 1a). The screen was performed against two Drosophila models of TDP-43 proteinopathy. Human TDP-43 was expressed with or without added ATXN2 bearing an intermediate polyglutamine repeat expansion of 32 glutamines (Q), an established risk factor for ALS [15, 39]. The disease-associated genes, as well as the RNAi transgenes, were expressed in the fly eye using the GAL4/UAS system [4]. 107 Drosophila genes that encode proteins with an RNA recognition motif (RRM) and have a direct human orthologue were targeted by RNAi. We focused on RRM containing genes because many of the genes associated with ALS contain such domains [21].

The screen identified a total of 22 modifiers (Fig. 1b, full screen results described in Additional file 1: Table S1). Of these, 3 showed strong and consistent effects between the TDP-43 and TDP-43/ATXN2-32Q models: Half pint (hfp, also known as pUF68) RNAi enhanced TDP-43 and TDP-43/ATXN2-32Q toxicity, while Ref1 and second mitotic wave missing (swm) RNAi suppressed the degenerative eye effect. We used a reporter gene, LacZ, to rule out whether these modifiers affected the GAL4/UAS expression system of the transgenes. This showed that RNAi to hfp increased and swm decreased the levels of β-galactosidase protein (Additional file 2: Figure S1A); importantly, Ref1 had no effect on the RNA or protein expression from the control LacZ gene (Fig. 1c, d). We confirmed efficient knockdown of Ref1 by the RNAi line (Additional file 2: Figure S1B). Ref1 was the strongest hit from the screen as it suppressed neurodegeneration caused by TDP-43 (Fig. 1c, d) and TDP-43/ATXN2-32Q (Fig. 1e), without an effect on expression of an unrelated control protein.

As ALYREF is a known mediator of RNA metabolism, we hypothesized that loss of Ref1 could alter the level of expression of the TDP-43 mRNA. Total RNA was extracted from fly heads co-expressing TDP-43. TDP-43 mRNA levels were determined using quantitative realtime polymerase chain reaction (qRT-PCR). Knockdown of Ref1 significantly reduced TDP-43 mRNA levels (Fig. 2a). Western immunoblot analysis was then used to determine the effects of Ref1 knockdown on TDP-43 protein levels. Consistent with its effects on TDP-43 mRNA, Ref1 RNAi reduced TDP-43 protein levels (Fig. 2b).

To gain an understanding if the effect of Ref1 knockdown was universal to all disease genes, we tested for effects of Ref1 depletion on expression of the disease gene Tau. Tau (encoded by the MAPT gene) is hyperphosphorylated and aggregates in a class of neurodegenerative diseases termed Tauopathies, which includes a subset of FTD cases lacking abnormal TDP-43 inclusions [47]. Thus, Tau represents an unrelated disease etiology as it does not co-occur with TDP-43 pathology. The reduced expression from the TDP-43 transgene in response to Ref1 RNAi was specific to this disease gene as Ref1 RNAi had no effect on mRNA expression (Fig. 2c) or protein expression (Fig. 2d) from a wild-type Tau transgene. These data indicated that Ref1 mitigated toxicity by selectively reducing levels of the disease TDP-43 mRNA.

ALYREF (human Ref1 protein homolog) was previously reported to bind G4C2 RNA [11, 19] and the presence of G4C2 expansions is found in TDP-43-associated ALS/FTD. Thus, we hypothesized that Ref1 may be able to also modify toxicity caused by expression of > 30 G4C2 repeats. Ref1 had previously been proposed as a modifier of G4C2-induced toxicity, causing accumulation of total mRNA within the nucleus when depleted [16]. However, Ref1 downregulation was not able to significantly modify toxicity associated with a shorter (G4C2)36 gene [19]. To further examine Ref1 as a potential modifier of expanded G4C2, we used a fly model expressing (G4C2)49 that induces neurodegeneration [6, 18, 26, 37]. Co-expression of Ref1 RNAi and expanded G4C2 resulted in suppressed degeneration in the fly eye (Fig. 3a). Importantly, this effect was consistent in both the external eye – seen by reduced pigment loss and recovered ommatidial organization – and in the internal retina tissue – seen by increased tissue integrity.

Given the effect of Ref1 RNAi on expression of the TDP-43 mRNA (see Fig. 2) and that recent evidence supports that ALYREF may play a more global role in transcription [5, 41, 49, 52, 58], we hypothesized that Ref1 may also alter expression of expanded G4C2. To assess the effects of Ref1 on expression, total RNA was extracted from fly heads co-expressing the G4C2 repeat transgene together with RNAi to Ref1 or a control. G4C2 RNA levels were determined using qRT-PCR [18]. Knockdown of Ref1 caused a significant reduction in the mRNA levels of G4C2 repeats (Fig. 3b).

G4C2 RNA can produce three dipeptide repeats (DPR) by non-AUG (RAN-) translation: glycine-arginine (GR), glycine-alanine (GA), glycine-proline (GP) [1, 34]. These DPR can form aggregates in ALS/FTD tissue. Of these, GR dipeptides have been consistently shown to be toxic in Drosophila with GA showing mild effects, and GP no toxicity [16, 36]. To determine if Ref1 knockdown effected GR protein levels, we utilized a Drosophila transgenic line that expresses a G4C2 repeat with a green fluorescent protein (GFP) tag in frame with the GR dipeptide [18]. Fluorescent imaging revealed that knockdown of Ref1 dramatically reduced GR-GFP accumulation (Fig. 3c). Blinded quantification of the GFP signal revealed a significant and consistent downregulation of the GR-GFP signal.

Overall, modification of G4C2-toxicity was consistent with a previous report [16], while our data further show that Ref1 loss alters the mRNA level of G4C2. Importantly, as Ref1 also altered expression of TDP-43 mRNA, it may serve as a unique target that can modify these two co-existing pathologies simultaneously.

As our data supported that knockdown of Ref1 modulates toxicity of TDP-43 and G4C2, we considered whether expression of endogenous Ref1 might be impacted by expression of these disease genes. To examine this, we expressed TDP-43 or a G4C2 repeat expansion in neurons of adult flies for 16d using the drug-inducible neuronal driver ElavGS, and then measured endogenous Ref1 mRNA levels in fly heads by qRT-PCR. Interestingly, endogenous Ref1 mRNA levels were significantly increased in animals expressing TDP-43 (Fig. 4a). Further, Ref1 mRNA levels were significantly increased upon expression of long and toxic (49 units), but not short and inert (8 units), G4C2 repeats (Fig. 4b). Taken together, these data suggest that Ref1 activity is upregulated by TDP-43 and toxic expanded G4C2, and that reduction of that activity by RNAi in Drosophila is beneficial.

Thus far, we found that Ref1 depletion in flies could suppress toxicity caused by expression of ALS/FTD disease genes TDP-43 and G4C2. Suppression was the result of reduced mRNA levels of these disease genes. Moreover, expression of TDP-43 and expanded G4C2 in the adult fly nervous system caused upregulation of endogenous Ref1. Given these findings, we were curious as to whether there may be an alteration in the expression of ALYREF, the human Ref1 orthologue, in human disease.

Immunofluorescence imaging for ALYREF protein was performed on lumbar spinal cord tissue sections to define any changes in localization and/or expression of ALYREF in disease (Fig. 5a). ALYREF is known to predominantly localize to the nucleus, although it can translocate to the cytoplasm as it accompanies exported nuclear mRNAs into the cytoplasm [3, 54, 57, 63]. Using two independent antibodies to ALYREF, we found that ALYREF localized to both the nucleus and cytoplasm in ALS motor neurons (Fig. 5b; patient information can be found in Additional file 3: Table S2). Importantly, ALYREF protein levels were significantly upregulated in ALS motor neurons compared to controls. Blinded quantification of ALYREF intensity revealed that upregulation of ALYREF was stronger in C9+ ALS cases, which have both TDP-43 pathology and the G4C2 repeat expansion present in C9orf72 [1, 13, 38, 42], compared to C9- ALS cases, which have only TDP-43 pathology [14, 32] (Fig. 5c). These data are consistent with the fly data showing that Ref1 (the orthologue of ALYREF in Drosophila) is upregulated upon expression of TDP-43 and G4C2. Moreover, these data support ALYREF dysregulation in human ALS/FTD.

Flies were grown on standard cornmeal molasses agar with dry yeast. Stock lines were maintained at 18 °C. Transgenic lines used in this study were: UAS-TDP-43/CyO; GMR-GAL4 (YH3)/TM6B. UAS-TDP-43(37 M), UAS-hATXN2-32Q (F26)/CyO; GMR-GAL4 (YH3)/TM6B and GMR-GAL4 (YH3)/TM3, Sb [25]. UAS-(G4C2)49, GMR-GAL4 (YH3)/TM6, Sb [18, 26, 37]. UAS-LDS-(G4C2)4,42,44[GR-GFP] [18]. UAS-Tau^WT [56]. RNAi lines from the Transgenic RNAi Project (TRiP) and mutant lines were obtained from the Bloomington Drosophila Stock Center. Additional RNAi lines were obtained from the Vienna Drosophila Resource Center. Complete list of RNAi lines used in this study is found in Additional file 4: Table S3.

For the genetic screen, virgin female flies were selected from each disease model line or from driver-only lines and were crossed to males harboring RNAi transgenes at 25 °C, under normal light/dark cycles. Male progeny of the appropriate genotypes were selected, aged to 3-5d at 25 °C. For external eye imaging, flies were anesthetized with ether for 10 min, placed on a glass slide and imaged with Leica Z16 APO. For internal eye morphology, male flies from the same crosses were fixed in Bouin’s solution (Sigma-Aldrich) for 120 h, embedded in paraffin, sectioned on a Leica RM2255 microtome and mounted on SuperFrost Plus slides (Fischer Scientific). Slides were dried overnight at room temperature, baked for 1 h at 56 °C, and paraffin was removed with Histoclear (National Diagnostics). Slides were mounted with coverslips using Cytoseal XYL (Thermo Scientific) and imaged using an upright Leica fluorescent microscope.

20 fly heads were homogenized in 50 μl LDS sample buffer (Invitrogen) including 5% beta-mercaptoethanol (Sigma-Aldrich). Samples were boiled at 95 °C for 5 min and centrifuged at 15,000 g for 5 min at 4 °C. The supernatant was collected and stored at − 20 °C until loaded on 4–12% Bis-Tris NuPAGE gels (Invitrogen) using 5 μl of sample per well. Gel electrophoresis was performed at 140 V for 70 min and the gels were blotted on a PVDF membrane using XCell II (Invitrogen) at 30 V for 1 h. Membranes were blocked in 3% bovine serum albumin in tris buffered saline with 0.1% Tween20 (TBST) for 30 min and incubated with primary antibodies in blocking buffer over-night at 4 °C. Following washes in TBST, membranes were incubated with HRP-conjugated secondary antibodies (Jackson Immunoresearch) at 1:10,000 for 2 h, washed and the luminescent signal was developed using ECL prime (Amersham) and detected with Amersham Imager 600. Primary antibodies: anti β-tubulin (CAT#E7, DSHB, 1:500), anti β-Galactosidase (CAT#Z378A, Promega, 1:2000), anti-TDP-43 (CAT#10782, Proteintech, 1:1000), anti Tau (CAT#A0024, Dako, 1:1000).

RNA was extracted using Trizol Reagent (ThermoFisher Scientific), according to the manufacturer’s instructions. RNA concentration was determined using Nanodrop (Nanodrop) and RNA quality was assessed using 1% agarose gel-electrophoresis. 400 ng RNA was used per reverse-transcription reaction using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) in a 20 μl total reaction volume using random primers. cDNA was then used as template for real-time qPCR using Fast SYBR Green Master Mix (Applied Biosystems). Real-time PCR was performed on the Applied Biosystems ViiA7 machine using 384-well format in technical duplicates. For each primer set, a serial dilution curve validated primer efficiency. Melting curve analysis confirmed the existence of one amplicon. Primers were designed using Primer3 (http://​bioinfo.​ut.​ee/​primer3-0.​4.​0/​primer3/​).

Primer sequences are listed in Additional file 5: Table S4.

For immuno-fluorescence, Superfrost slides (Fisher Scientific) with paraffin sections were deparaffinized in xylene, 100, 90 and 70% ethanol. Following a brief rinse in water, antigen retrieval was performed by boiling the samples for 10 min in citric buffer pH 6 (10 mM citric acid, pH 6). Slides were cooled to room temperature, rinsed with water and incubated in blocking buffer for 20 min (Tris buffered saline (TBS) with 2% bovine serum). Sections were circled with a liquid blocker PAP pen (Daido Sangyo) and primary antibodies in blocking buffer were incubated overnight at 4 °C. Following 3 washes in TBS, slides were blocked for 5 min and incubated with secondary antibodies for 2 h at room-temperature. Slides were washed 3 times in TBS. To quench autofluorescence, slides were washed for 1 min in 70% ethanol, incubated for 1 min in Sudan Black (0.3% in 70% ethanol), and washed 3 times in 70% ethanol. Final TBS wash was followed by 1 μg/ml DAPI to stain DNA, slides were rinsed in water, mounted with anti-fade mounting media (20 mM Tris pH 8.0, 0.5% N-propyl gallate, 80% glycerol), and sealed with clear nail polish. Quantification of fluorescent signal was performed with Fiji [45]. Imaging and quantification were performed blinded to disease status. Primary antibodies were used at 1:200 dilution: anti-ALYREF (mouse monoclonal, CAT# ab6141, abcam), anti-ALYREF (rabbit polyclonal, CAT# ab202894, abcam). Secondary antibodies were used at 1:200 dilution: anti-Rabbit IgG Alexa Fluor 594 (#A-11012, 1:200, Invitrogen). Details of human samples are described in Additional file 3: Table S2. Donor spinal cord samples following neuropathological evaluation were selected from the brain bank at the Center for Neurodegenerative Disease Research at the University of Pennsylvania [51]. Phosphorylated TDP-43 deposits were evaluated using the pS409/410 antibody (mAb, 1:500) [40]. Controls were defined as subjects who were cognitively normal and did not meet the threshold for a neurodegenerative or vascular dementia diagnosis during the neuropathological examination. Informed consent for autopsy was obtained for all patients from their next of kin.

Statistical analysis was performed using Prism (Version 6, GraphPad). One-way ANOVA with Dunnett’s or Tukey’s multiple comparisons test or two tailed Student’s t-test were used as appropriate, with significance level set at 0.05.