eIF4B and eIF4H mediate GR production from expanded G4C2 in a Drosophila model for C9orf72-associated ALS

Abstract

The discovery of an expanded (GGGGCC)n repeat (termed G4C2) within the first intron of C9orf72 in familial ALS/FTD has led to a number of studies showing that the aberrant expression of G4C2 RNA can produce toxic dipeptides through repeat-associated non-AUG (RAN-) translation. To reveal canonical translation factors that impact this process, an unbiased loss-of-function screen was performed in a G4C2 fly model that maintained the upstream intronic sequence of the human gene and contained a GFP tag in the GR reading frame. 11 of 48 translation factors were identified that impact production of the GR-GFP protein. Further investigations into two of these, eIF4B and eIF4H, revealed that downregulation of these factors reduced toxicity caused by the expression of expanded G4C2 and reduced production of toxic GR dipeptides from G4C2 transcripts. In patient-derived cells and in post-mortem tissue from ALS/FTD patients, eIF4H was found to be downregulated in cases harboring the G4C2 mutation compared to patients lacking the mutation and healthy individuals. Overall, these data define eIF4B and eIF4H as disease modifiers whose activity is important for RAN-translation of the GR peptide from G4C2-transcripts.

Keywords: Amyotrophic lateral sclerosis (ALS) (Lou Gehrig disease); C9orf72; Drosophila; Neurodegeneration; RAN-translation; eIF4B; eIF4H.

Fig. 1

Expanded G4C2 transgenes produce GFP-tagged GR. a. A new transgenic (G4C2)n model was developed to look at RAN-translation of the GR reading frame. A “leader” sequence (LDS) was added 5′ of the repeat: 114 bp of intronic sequence found upstream of the repeat in patient samples. The GR reading frame has an in-frame GFP coding sequence 3′ of the repeat that lacks an ATG initiation. b. To define the number of repeats inserted into genomes of w¹¹¹⁸ transgenic flies, PCR reactions were developed that amplified the repeat and its flanking region. The number of repeats were calculated from PCR product lengths measured on a Bioanalyzer and agarose gel. Shown: the maximum number of repeats found in the control (CTRL) or expanded (EXP) G4C2 fly lines; individual data points from 2 independent DNA preps with mean. c. qPCR analysis for RNA levels between control and expanded G4C2 fly lines. Shown: individual data points with mean ± SD. Statistics: unpaired student t-test, p-value **** < 0.0001. d. Western immunoblots confirmed GR is produced and successfully tagged with GFP in EXP-G4C2 flies. No GR/GFP was detected in control G4C2 flies, even with overexposure (Additional file 1: Figure S1). Uncropped westerns (Additional file 1: Figure S4) e. External and internal eye analysis in animals expressing G4C2 or control (DSRED) transgenes using GMR-GAL4. Degeneration was seen only in LDS-(G4C2)_EXP animals: external pigment loss and reduced integrity of internal retina tissue. f. External eye imaging for fluorescence caused by transgene expression. Positive (DSRED) control flies show uniform, diffuse signal. Control G4C2 flies show no signal, even with increased exposure (data not shown). Expanded G4C2 flies show GFP puncta. Shown (e-f): representative images while all conditions were tested 2+ times. For full genotypes see Additional file 7: Table S1

Fig. 2

A screen of translation factors reveals those important for expression of GR from (G4C2)_EXP. a. To identify canonical translation factors that may be involved in RAN translation of G4C2 in the GR reading frame, a loss-of-function (LOF) based screen was designed utilizing previously developed RNAi [57, 60] or LOF mutant fly lines [6, 7, 80, 81] targeting 48 of 56 (86%) known translation factors [50]. Individual translation factors were downregulated in animals expressing LDS-(G4C2)_EXP and any that altered the external eye phenotype and/or GR-GFP levels were defined (Step 1). These 28 LOF lines were further tested in (GR)₃₆ expressing animals, defining 6 that acted similarly on GR-associated toxicity (Step 2). Additional quality control experiments excluded LOF lines that altered expression from a control (LacZ) transgene by western immunoblot and/or altered a WT eye morphology when expressed alone, as described [13, 26, 41] (Step 3). b. Summary of screen results. Overall, 11 translation factors were identified as candidate RAN translation factors as their depletion reduced GR-GFP levels. Overall, excluded LOF lines either: altered toxicity of G4C2 flies but did not alter GR-GFP levels, similarly altered toxicity in a non-G4C2, GR fly model arguing that these acted downstream of GR production, had no effect on G4C2 toxicity or GR-GFP levels, or were “unspecific” modifiers identified by quality control experiments. Shown: representative images while all RNAi were tested 2+ times for effects under each condition. Details on LOF lines used and complete results with each line can be found in Additional file 2: Table S2. For full genotypes and RNAi lines see Additional file 7: Table S1

Fig. 3

Analysis of eIF4B and eIF4H1 RNAi in control flies. a. RNA levels produced from eIF4B or eIF4H1 were assessed by qPCR in flies ubiquitously expressing RNAi (by Daughterless-GAL4). Statistics: one-way ANOVAs with Tukey’s multiple comparison correction, p-values **** < 0.0001, *** < 0.001, ** < 0.01, * < 0.05, no significance > 0.05. Shown: individual data points from 2 independent experiments with mean ± SEM. b. Viability studies in Drosophila reveal that RNAi-depletion of eIF4B or eIF4H1 does not significantly alter the number of adult flies expected to eclose. Shown: ratio of progeny from two individual crosses with RNAi compared to sibling animals with the balancer chromosome that reach adulthood (1-2d adult animals). Comparing the # progeny that eclose from a single vial compensates for differences in mating variability, fertilized eggs laid, among other variables. However, we note that the presence of the balancer chromosome could potentially cause mild sub-viability to adulthood. Crosses: RNAi/CyO x Da-GAL4 (III); counted progeny: RNAi/+; Da-GAL4/+ and CyO/+; Da-GAL4/+. RNAi lines: control (JF01355), eIF4B RNAi (HMS04503), eIF4H1 RNAi (HMS04504). For full genotypes see Additional file 7: Table S1

Fig. 4

Depletion of eIF4B and eIF4H1 selectively suppresses LDS-(G4C2)_EXP associated toxicity. a. Using GMR-GAL4, RNAi-mediated depletion of eIF4B or eIF4H1 in LDS-(G4C2)_EXP expressing flies results in reduced toxicity in both the external and internal eye: seen externally by recovered pigment and ommatidial structure, seen internally by recovered retinal tissue integrity. b. Blinded quantification of internal retina tissue was done by measuring the total surface area of tissue present and by measuring the depth of the tissue at the position where the optic chiasm occurs. n = 9–10 animals per genotype. c. eIF4B or eIF4H1 RNAi was expressed in (GR)36 flies (GMR-GAL4) and effects on GR-associated toxicity were observed in the external and internal eye. d. Blinded quantification of internal retina tissue. n = 5–9 animals per genotype. e. eIF4B or eIF4H1 RNAi was expressed in control flies (GMR-GAL4) and effects on the normal eye were observed externally and internally. f. Blinded quantification of internal retina tissue. n = 4 animals per genotype. For graphs, shown are individual data points representing 1 animal with mean ± SD. Statistics: one-way ANOVAs with Tukey’s multiple comparison correction, p-values **** < 0.0001, *** < 0.001, ** < 0.01, * < 0.05, no significance > 0.05. RNAi lines: control (JF01355), eIF4B (HMS04503), eIF4H1 (HMS04504). For full genotypes see Additional file 7: Table S1

Fig. 5

eIF4B and eIF4H1 RNAi selectively reduce GR-GFP levels produced from LDS-(G4C2)_EXP. a. Fluorescence imaging in LDS-(G4C2)_EXP expressing flies shows that depletion of eIF4B or eIF4H by RNAi results in reduced GR-GFP levels (GMR-GAL4). b. Blinded quantification of GR-GFP signal in LDS-(G4C2)_EXP animals relative to the signal in control RNAi animals. Analysis of GR-GFP puncta number and size are shown in black. Total GR-GFP signal is shown in grey. n = 6–7 animals per genotype. Further, qPCR was used to quantify RNA levels in LDS-(G4C2)_EXP flies co-expressing control, eIF4B, or eIF4H1 RNAi, shown in light grey. c. A control fluorescent protein, DsRed, was similarly expressed in the fly eye with RNAi to control, eIF4B, or eIF4H. d. Blinded quantification of DsRed fluorescence shows no effect by RNAi. n = 6 animals per genotype. e. A representative western immunoblot image for β-Galactosidase in LacZ flies co-expressing control, eIF4B, or eIF4H1 RNAi. Uncropped westerns (Additional file 1: Figure S4). Blinded quantification of β-Galactosidase western immunoblots normalized to the loading control, Tubulin. For graphs, shown are individual data points from 2 independent assays with mean ± SEM. Statistics: one-way ANOVAs with Tukey’s multiple comparison correction, p-values **** < 0.0001, *** < 0.001, ** < 0.01, * < 0.05, no significance > 0.05. RNAi lines: control (JF01355), eIF4B (HMS04503), eIF4H1 (HMS04504). For full genotypes see Additional file 7: Table S1

Fig. 6

EIF4H is downregulated in C9+ ALS/FTD. a. Western immunoblots were used to define changes in eIF4B or eIF4H protein levels from 5 independent control or 4 independent C9+ derived fibroblast cell lines. Data are relative to controls. Quantification of total protein was done after normalizing to loading, using GAPDH. Phospho-eIF4B quantification was further normalized to total eIF4B. Statistics: unpaired student t-tests. Shown: each data point represents 1 cell line with mean ± SEM; the mean data from 2 independent protein preparations is shown per line. b. RNA levels of EIF4B or EIF4H were assessed by qPCR in human cerebellar tissue from healthy individuals or ALS/FTD patients with (C9+) or without (C9-) the G4C2 expansion in C9orf72. n = 22 (healthy), 46 (C9- ALS/FTD), 66 (C9+ ALS/FTD). Statistics: one-way ANOVAs with Dunn’s multiple comparison correction. Shown: individual data points representing 1 individual with mean ± SEM. Cell line details: Additional file 3: Table S3. Patient details: Additional file 4: Table S4. p-values **** < 0.0001, *** < 0.001, ** < 0.01, * < 0.05, no significance > 0.05

Fig. 7

Model comparing potential G4C2 RAN-translation mechanisms and canonical translation. a. Ternary complex formation requires eIF5-mediated exchange of GDP to GTP on eIF2 complex (includes eIFs 2α, 2β, 2γ). eIF2α is highly regulated during stress and is reported to mediate G4C2 translation [12, 27]. eIF2β and eIF5 were identified as modifiers in this study. b. In normal translation, the formation of the 43S pre-initiation complex (PIC) involves the joining of a number of factors, including Ternary complex (described in a) and eIFs 1, 1A, 3, and 5. c. A minimal PIC complex may potentially mediate RAN-translation [1, 42, 76, 86]. d. mRNA transcripts are recognized by the eIF4F complex, includes eIFs 4E, 4G, 4A. All of these have been defined as G4C2 translation factors arguing that G4C2 RAN-translation is cap-dependent [12, 84]. eIF4E recognizes the 5-prime m⁷G cap on mRNAs [10, 77, 78]; notably, 4 of 6 eIF4E components were identified in our screen. eIF4A is recruited by eIF4E to mRNA transcripts (via the scaffold protein eIF4G). mRNA is then unwound by eIF4A, an activity that is significantly promoted by eIF4B or eIF4H, identified herein [24, 68, 70, 82, 91]. This action allows for the formation of the 48S scanning complex. e. In canonical translation, the 48S scanning complex moves down a transcript until identifying an AUG start codon. A CUG codon in the LDS sequence upstream of G4C2 may function as a start codon in the GA-reading frame [27, 84]. Frame-shifting could allow for translation of the GR and GP from this codon. Candidate RAN translation factors eIF5B, and potentially eIF5, mediate ribosome scanning, start codon recognition, and translation activation [10, 45, 61]. We note that mechanisms underlying RAN-translation are still relatively unknown. This model is based on current literature and canonical functions of translation factors