Amyotrophic Lateral Sclerosis Modifiers in Drosophila Reveal the Phospholipase D Pathway as a Potential Therapeutic Target

Abstract

Amyotrophic lateral sclerosis (ALS), commonly known as Lou Gehrig's disease, is a devastating neurodegenerative disorder lacking effective treatments. ALS pathology is linked to mutations in >20 different genes indicating a complex underlying genetic architecture that is effectively unknown. Here, in an attempt to identify genes and pathways for potential therapeutic intervention and explore the genetic circuitry underlying Drosophila models of ALS, we carry out two independent genome-wide screens for modifiers of degenerative phenotypes associated with the expression of transgenic constructs carrying familial ALS-causing alleles of FUS (hFUS^R521C) and TDP-43 (hTDP-43^M337V). We uncover a complex array of genes affecting either or both of the two strains, and investigate their activities in additional ALS models. Our studies indicate the pathway that governs phospholipase D activity as a major modifier of ALS-related phenotypes, a notion supported by data we generated in mice and others collected in humans.

Keywords: Amyotrophic lateral sclerosis (ALS); C9ORF72; FUS; TDP43; phospholipase D (PLD).

Figure 1

Screening strategy and primary screen validation. (A) Schematic of primary screens for Exelixis insertions that alter degenerative eye phenotypes associated with w; GMR-GAL4; UAS-hFUS^R521C (GMR-hFUS^R521C) and w; GMR-GAL4, UAS-hTDP-43^M337V (GMR-hTDP-43^M337V). These transgenic Drosophila models display photoreceptor degeneration/rough-eye phenotypes that are fully penetrant and dosage-sensitive (Ritson et al. 2010; Lanson et al. 2011; Periz et al. 2015). For the primary screen, we generated F1 individuals carrying an Exelixis insertion in trans with either GMR-hFUSR^521C or GMR-hTDP-43^M337V, which were scored for enhancement or suppression of the eye degeneration phenotype. All primary screen positive inserts were retested in a validation screen using an identical crossing scheme to confirm the initially observed interaction(s). (B) Control GMR-GAL4 heterozygous individual. The eyes of (C) GMR-hFUSR^521C and (H) GMR-hTDP-43^M337V heterozygous animals displaying degenerative, pigmentation, and rough-eye phenotypes. (D–G) GMR-hFUS^R521C and (I–L) GMR-hTDP-43^M337V were modulated by Exelixis inserts disrupting genes with known associations to ALS: (D and I) dSenataxin^f05408, (E and J) discs overgrown^d06510, (F and K) Hsc70Cb^d03562, and an (G and L) GAL4-inducible RNAi allele Ask1³²⁴⁶⁴. All eyes shown are representative images from individual females. Exelixis stock IDs and Drosophila gene symbols are listed, Ask1^RNAi refers to Bloomington Strain ID (BSID) 32464.

Figure 2

Screen results. (A) Venn diagrams showing the number of validated overlapping inserts with Drosophila gene assignments recovered in the screens. In total, 637 GMR-hFUS^R521C and 553 GMR-hTDP-43^M337V modifying insertions (enhancers and suppressors) were recovered and validated in each screen, with 432 insertions recovered in both screens. For GMR-hFUS^R521C 277 suppressors and 360 enhancers were recovered; for GMR-hTDP-43^M337V the totals were 249 and 304 suppressors and enhancers, respectively, which includes 173 overlapping suppressors and 259 overlapping enhancers. We note that genes recovered in both screens resulting in opposite modification phenotypes were not included in the suppressors and enhancers Venn diagrams. We also note that Table S1 includes all inserts recovered, not only those with clear Drosophila gene assignments in the Exelixis Collection. Panels B–S show examples of novel modifiers identified in our screens. Female eyes trans-heterozygous GMR-hFUS^R521C (C–O) and with GMR-hTDP-43^M337V(H–S) and enhancing or suppressing mutations are shown. Drosophila gene symbols and Exelixis Stock IDs are listed. Controls (B and G) demonstrate the eye degeneration phenotype. (C and H) hdc^d10800 suppresses both screening strains, (D) Pka-R2^d02258 has no effect on GMR-hFUS^R521C, but (I) suppresses GMR-hTDP-43^M337V, (E and J) Strip^e04482 enhances both screening strains, (F) sas^d07239 suppresses GMR-hFUS^R521C, and (K) enhances GMR-hTDP-43^M337V. (L–S) Mutations in different RBPs affect the GMR-hFUS^R521C and GMR-hTDP-43^M337V eye phenotypes. (L and P) Atx-1^f01201 suppresses both screening strains, (M and Q) pum^d04225 and (N and R) orb^d06989 enhance GMR-hFUS^R521C and weakly suppress GMR-hTDP-43^M337V, and (O and S) glo^f02674 suppresses GMR-hFUS^R521C and enhances GMR-hTDP-43^M337V.

Figure 3

C9ORF72 GGGGCC hexanucleotide repeat model of progressive neurodegeneration. The GAL4-inducible c9orf72(G4C2)₃₀ transgenic construct was assessed for a progressive neurodegenerative phenotype using the GMR-GAL4 driver over a period of 6 weeks. Neurodegeneration was scored as the presence of black necrotic tissue on the cuticle of the eye. Within a given genotype, the penetrance of neurodegeneration was determined as the fraction of individuals exhibiting black spots within the entire population at multiple time points. Two crosses per genotype were examined and averaged to determine the penetrance of the phenotype. The presence of a single spot on a single eye was considered positive, regardless of the magnitude of the spot(s). (A–D) Representative eye images of aged GMR-GAL4 individuals display no cuticular photoreceptor degeneration as determined by the presence of black necrotic spots over a period of 6 weeks. (E–H) Representative images of aged GMR-GAL4, UAS-c9orf72(G4C2)₃₀/+ (GMR-c9orf72₃₀) individuals. Eyes were photographed at day 1 (shortly after hatching), and after 1, 3, and 6 weeks. We note that the darkening of the eye color is an aging effect and does not reflect degeneration. (I) Histogram representation of percent GMR-c9orf72₃₀/+ individuals (control) displaying the degenerative phenotype at week 1 (blue), week 3 (orange), and week 6 (gray) showing an increase in penetrance of the degenerative phenotype with age. Several genes with established links to ALS are shown to dominantly suppress the GMR-c9orf72₃₀/+ progressive degeneration validating GMR-c9orf72₃₀ as an ALS-related model. Gene symbols and the Bloomington Stock IDs (BSID) are provided. dGLE1^RNAi: BSID – 52888, HDAC6^RNAi: BSID – 31053, futsch^EP1419: BSID – 10571, Ask1^RNAi: BSID – 35331 and Hsc70Cb^d03562 is an Exelixis Hsc70Cb allele. The animals of the resulting genotypes were examined: GMR-c9orf72₃₀/+ (control), GMR-c9orf72₃₀/dGLE1⁵²⁸⁸⁸ (dGLE1^RNAi), GMR-c9orf72₃₀/+; HDAC6³¹⁰⁵³/+ (HDAC6^RNAi), futsch^EP1419/+; GMR-c9orf72₃₀/+ (futsch^EP1419), GMR-c9orf72₃₀/+; Hsc70Cb^d03562/+ (Hsc70Cb^d03562), and GMR-c9orf72₃₀/+; Ask1³⁵³³¹/+ (Ask1^RNAi).

Figure 4

Drosophila TDP-43 ALS model. (A–F) Eyes from animals expressing GMR-GAL4 control (A) compared to GMR-dTDP-43^WT (B) and GMR-dTDP-43^mNLS (C). Expression of each construct results in ommatidia loss and a glossy appearance. GMR-GAL4-driven expression of dTDP-43^N493D results in larval lethality. (D–F) Cytoplasmic aggregates are observed in eye discs from individuals with GMR-GAL4-driven (D) wild-type dTDP-43^WT, (E) dTDP-43^N493D, and (F) dTDP-43^mNLS and stained with an antibody directed against human TDP-43 (red); nuclei are stained with DAPI (blue). We note that in control animals relatively low levels of endogenous dTDP-43 expression was detected using this antibody (Figure S2A). (G–O) OK371-GAL4 drives expression of UAS containing transgenic constructs in larval motor neurons; the OK371-GAL4 driver strain carries UAS-CD8-GFP, which labels the cell membrane of transgene-expressing cells with GFP (green). As above, ectopic dTDP-43 protein is detected with anti-TDP-43 (red) and nuclei are stained with DAPI (blue). (G–I) UAS-dTDP-43^WT and (J–L) UAS-dTDP-43^N493D show large cytoplasmic aggregated forms of dTDP-43, with little evidence of more broadly distributed cytoplasmic distributions, whereas (M–O) UAS-dTDP-43^mNLS exhibits dTDP-43 cytoplasmic mislocalization but appears to produce fewer aggregates and a more diffuse cytoplasmic pattern.

Figure 5

Modifier effects on dTDP-43 transgenic construct expression in the NMJ. (A) Table summarizing the observed phenotypes using two different GAL4 motor neuron drivers (OK371-GAL4 and OK6-GAL4) to direct expression of the following UAS containing transgenic constructs inserted into the ZH-86Fb attB third chromosome insertion site: dTDP-43^WT, dTDP-43^mNLS, and dTDP-43^N493D. (B–G) Qualitative morphological effects on larval NMJs of the following genotypes: (B) OK371-GAL4, (C) OK371-GAL4/+; UAS-dTDP-43^WT, (D) OK371-GAL4/+; UAS-dTDP-43^mNLS/+, (E) OK371-GAL4/+; UAS-dTDP-43^N493D/+, (F) OK371-GAL4/+; UAS-dTDP-43^N493D/SF2³²³⁶⁷, and (G) OK371-GAL4/+; UAS-dTDP-43^N493D/lilli²⁶³¹⁴. (H) Histogram representation of the quantification of the average bouton numbers per muscle in individuals from B–G. All genotypes listed are in an OK371-GAL4/+ background: OK371 are control OK-371-GAL4 heterozygous individuals (gray), WT (light blue), mNLS (red), and N493D (yellow) correspond to the dTDP-43 transgenic constructs, while SF2/N493D (green), lilli/N493D (blue), and klp98A/N493D (magenta) correspond to the SF2³²³⁶⁷, lilli²⁶³¹⁴, and klp98A^c05668 alleles in the background OK371-GAL4/+; UAS-dTDP-43^N493D. The dTDP-43^N493D strain displayed significant differences in NMJ bouton counts when compared to OK371-GAL4 (**** P < 0.001), OK371-GAL4/+; UAS-dTDP-43^WT (**** P < 0.001), and OK371-GAL4/+; UAS-dTDP-43^mNLS (**** P < 0.001). dTDP-43^mNLS and dTDP-43^WT bouton numbers were not significantly different from dTDP-43^WT. SF2 (* P < 0.0031) and lilli (**** P < 0.001) alleles caused substantial improvement of NMJ morphology in comparison to OK371-dTDP-43^N493D, while klp98A did not. Quantifications were done manually at the confocal microscope and statistical significance was determined by using an unpaired parametric t-test with Prism software. NMJ preparations were stained with anti-HRP (green) and anti-discs large (Dlg) (red) to mark pre- and postsynaptic structures, respectively, and muscle nuclei were labeled with DAPI. The asterisks above the lines correspond to the degree of significance between the denoted genotypes. The more asterisks, the smaller the P-value.

Figure 6

dPLD effects on GMR-hFUS^R521C, GMR-hTDP-43^M337V, and GMR-c9orf72(G4C2)₃₀ phenotypes. (A–D) RNAi-induced dPld reduction suppressed GMR-hFUS^R521C and GMR-hTDP-43^M337V phenotypes. Representative eye images of individuals carrying one copy of (A) GMR-hTDP-43^M337V in trans with (B) UAS-dPld³²⁸³⁹, or (C) GMR-hFUSR^521C in trans with (D) UAS-dPld³²⁸³⁹. dPld³²⁸³⁹ expresses an RNAi directed against dPld. (E) Histogram representation of percentage of GMR-c9orf72(G4C2)₃₀ individuals displaying the degenerative phenotype at week 1 (blue), week 3 (orange), and week 6 (gray). GMR-GAL4-directed expression of UAS-dPld³²⁸³⁹ results in suppression of the c9orf72(G4C2)₃₀-dependent neurodegenerative phenotype at all three time points. (F–H) Effects of motoneuron-driven dPLD on third instar larval NMJ morphology. Representative images of third instar larval NMJs from (F) control OK371-GAL4 as well as loss [(G) UAS-dPld³²⁸³⁹] and gain [(H) UAS-dPld13] (Raghu et al. 2009) of dPld function in OK371-GAL4/+ background are shown. (I) Histogram representation of quantification of average bouton numbers per muscle in OK371-GAL4 (OK371), OK371-GAL4/+ UAS-dPld³²³⁸⁹/+ (dPLD.RNAi, gray), OK371-GAL4/UAS-dPld13 (UAS.dPLD, red), OK371-GAL4, UAS-dTDP-43^N493D (N493D, yellow), and OK371-GAL4, UAS-dTDP-43^N493D/UAS-dPld³²⁸³⁹ (N493D/dPLD.RNAi, green) individuals. Loss of dPld has no effect on OK371-GAL4 bouton number, whereas there is a statistically significant decrease (**** P < 0.0001) in bouton number when dPLD is ectopically expressed. Quantifications were done manually at the confocal microscope and statistical significance was determined by using an unpaired parametric t-test with Prism software. NMJ preparations were stained with anti-HRP (green) and anti-discs large (Dlg) (red) to mark pre- and postsynaptic structures, respectively, and muscle nuclei were labeled with DAPI. The more asterisks, the smaller the P-value.

Figure 7

dPLD1 pathway elements modify Drosophila ALS models. (A–H) Representative eye images of individuals containing carrying the (A) GMR-hTDP-43^M337V transgenes in trans with (B) UAS-Rgl^d03208, (C) UAS-Rala³⁴³⁷⁵ (Rala^RNAi), and (D) UAS-Rala³²⁰⁹⁴ (Rala^DN) and the (E) GMR-hFUS^R521C transgenes in trans with (F) UAS-Rgl^d03208, (G) UAS-Rala³⁴³⁷⁵ (Rala^RNAi), and (H) UAS-Rala³²⁰⁹⁴ (Rala^DN). Both FUS and the TDP-43 phenotypes were suppressed by Rgl^d03208 and Rala³⁴³⁷⁵ (Rala^RNAi). (I) Histogram representation of percentage of individuals displaying the degenerative phenotype at week 1, week 3, and week 6 for (I) GMR-c9orf72(G4C2)₃₀/+ control (blue), (J) UAS-Rala³⁴³⁷⁵/+; c9orf72(G4C2)₃₀/+ (Rala^RNAi) (orange), and UAS-Rala³²⁰⁹⁴/+; c9orf72(G4C2)₃₀/+ (Rala^DN) (gray). Both Rala alleles strongly suppressed the penetrance of the degenerative c9orf72 phenotype. (J–M) Motor-neuron-directed reduction of Rgl affects OK371-GAL4; UAS-dTDP-43^N493D/+ third instar larval NMJ morphology: representative images of control NMJs from (J) OK371-GAL4 (OK371/+) and (K) OK371-GAL4; UAS-dTDP-43^N493D/+ individuals. (L) The OK371-GAL4; UAS-dTDP-43^N493D/+ NMJ phenotype (K) appears to be qualitatively rescued by RNAi-induced reduction of Rgl (UAS-Rgl²⁸⁹³⁸) (L). (M) Quantification of average bouton numbers per muscle in individuals of the OK371-GAL4 (OK371), OK371-GAL4; UAS-dTDP-43^N493D/+ (N493D) and OK371-GAL4; UAS-dTDP-43^N493D/UAS-Rgl²⁹³⁹⁸ (N493D/Rgl.RNAi) genotypes. RNAi-induced reduction of Rgl results in a statistically significant increase in the number of boutons (**** P < 0.001). Quantifications were done manually at the confocal microscope and statistical significance was determined by using an unpaired parametric t-test with Prism software. NMJ preparations were stained with anti-HRP (green) and anti-discs large (Dlg) (red) to mark pre- and postsynaptic structures, respectively, and muscle nuclei were labeled with DAPI.

Figure 8

Effects of genetic deletion of PLD in SOD1^G93A Mice. Survival, weight, and behavior were measured every 10 days starting from P 100 (post natal day 100). (A) Elimination of PLD1, PLD2, or PLD1/2 combined had no effect on survival of SOD1^G93A transgenic animals. (WT (n = 12), SOD1^G93A (n = 16), SOD1^G93A; PLD1−/− (n = 20), SOD1^G93A; PLD2−/− (n = 20) and SOD1^G93A; PLD1−/− and PLD2−/− (n = 20) mouse strains). (B) Starting from P 100, mice were weighed every 10 days. PLD1 and/or PLD2 elimination had no effect on the weight loss observed in the SOD1^G93A animals. (C and E) Fore limb and hind limb grip strength test measured at P 120 and 130 showed significant improvement of both fore limb and hind limb strength in SOD1^G93A; PLD1−/− strains, compared to SOD1^G93A controls (* P < 0.05, two-way ANOVA, followed by Tukey’s multiple comparisons test). In addition, SOD1^G93A; PLD2−/− mice had greater fore limb strength and SOD1^G93A; PLD1 and PLD2 greater hind limb strength at P 120, compared to the SOD1^G93A mice (* P < 0.05) (n = 10 in SOD1 group and n = 20 in other groups). (D and F) Percentage change in grip strength from P 100 to P 120 (P 100 value − P 120 value)/P 100 value. Consistent with data in C and E, PLD knockout had mild beneficial effects in terms of fore and hind limb strength. (G and H) Inverted grip strength of the SOD1^G93A; PLD1−/− and SOD1^G93A; PLD2−/− mice is significantly improved at P 120 and P 130, respectively, compared to the SOD1^G93A control animals (* P < 0.05 compared to SOD1^G93A group, two-way ANOVA, followed by Tukey’s multiple comparisons test) (n = 10 in SOD1^G93A group and n = 20 in other groups), meanwhile the percentage change was also analyzed and showed similar results.

Figure 9

Multiple elements of the PLD1 pathway modify ALS-related phenotypes. Shown is a schematic depicting biochemical relationships leading to Pld production (Foster and Xu 2003). Red arrows point to pathway components identified as modifiers in our GMR-hFUS^R521C and/or GMR-hTDP-43^M337V screens. Green arrows point to elements reported to be upregulated in patients with early-onset, but not late-onset, ALS. In both the primary and validation screens six different Drosophila orthologs of components of the PLD1 pathway were recovered as modifiers of GMR-hFUS^R521C and GMR-hTDP-43^M337V: dPld, ArfGAP3, Rgl, Ras85D, Plc21C, and Rho1, which correspond to the human genes, PLD1, ARFGAP3, RALGDS, KRAS, PLCB1, and RHOA.