Fragile X premutation rCGG repeats impair synaptic growth and synaptic transmission at Drosophila larval neuromuscular junction

Abstract

Fragile X-associated tremor/ataxia syndrome (FXTAS) is a late-onset neurodegenerative disease that develops in some premutation (PM) carriers of the FMR1 gene with alleles bearing 55–200 CGG repeats. The discovery of a broad spectrum of clinical and cell-developmental abnormalities among PM carriers with or without FXTAS and in model systems suggests that neurodegeneration seen in FXTAS could be the inevitable end-result of pathophysiological processes set during early development. Hence, it is imperative to trace early PM-induced pathological abnormalities. Previous studies have shown that transgenic Drosophila carrying PM-length CGG repeats are sufficient to cause neurodegeneration. Here, we used the same transgenic model to understand the effect of CGG repeats on the structure and function of the developing nervous system. We show that presynaptic expression of CGG repeats restricts synaptic growth, reduces the number of synaptic boutons, leads to aberrant presynaptic varicosities, and impairs synaptic transmission at the larval neuromuscular junctions. The postsynaptic analysis shows that both glutamate receptors and subsynaptic reticulum proteins were normal. However, a high percentage of boutons show a reduced density of Bruchpilot protein, a key component of presynaptic active zones required for vesicle release. The electrophysiological analysis shows a significant reduction in quantal content, a measure of total synaptic vesicles released per excitation potential. Together, these findings suggest that synapse perturbation caused by riboCGG (rCGG) repeats mediates presynaptically during larval neuromuscular junction development. We also suggest that the stress-activated c-Jun N-terminal kinase protein Basket and CIDE-N protein Drep-2 positively mediate Bruchpilot active zone defects caused by rCGG repeats.

Introduction

The fragile X mental retardation 1 (FMR1) gene normally harbors a highly polymorphic trinucleotide repeat sequence (CGG) within its 5′ untranslated region (5′ UTR). The normal allele of the FMR1 gene typically has 5–40 CGG repeats. Abnormal alleles include the full mutation (>200 CGG repeats), premutation (PM) (55–200 CGG repeats) and gray zone mutation (45–54 CGG repeats). Carriers of full mutation develop fragile X syndrome (FXS; OMIM: 300624), the most common inherited form of neurodevelopment and intellectual disability (ID) disorder, occurring in 1 in 4000 to 1 in 7000 people (1–4). On the other hand, PM carriers account for a variety of phenotypes that are found frequently in the population, with an estimated prevalence of 1:259 in females and 1:813 in males (5,6). A proportion of these PM carriers, about 40% of males and 16% of females develop a progressive neurodegenerative disorder termed fragile X-associated tremor/ataxia syndrome (FXTAS; OMIM: 300623) usually after the fifth decade of life (7,8). Clinically, FXTAS presents with intention tremor, gait ataxia and other features including parkinsonism, cognitive defects, brain atrophy and white matter abnormalities on MRI (7,9). Neuropathologically, FXTAS is distinguished by the presence of large eosinophilic nuclear inclusions in neurons, astrocytes, in the spinal column and peripheral tissues (10–12). These inclusions contain expanded FMR1 messenger RNA (mRNA), ubiquitin and several other proteins that aggravate cellular dysregulation (13).

One unique molecular signature of the fragile X PM allele is that the level of FMR1 mRNA is significantly elevated; however, FMR1 encoded protein (FMRP) paradoxically remains relatively unchanged. Therefore, the neurodegenerative phenotypes associated with FXTAS are suspected of being caused by a gain of function in fragile X PM riboCGG (rCGG) repeat RNAs (14–17). Model systems developed to mimic the molecular and cellular alterations together with the clinical symptoms of FXTAS have proposed several mechanisms of pathogenicity that include RNA-mediated toxicity and repeat-associated non-AUG (RAN) mediated proteotoxicity (18–23). The RNA gain-of-function, however, seems to play a major role in rCGG-mediated pathogenesis (18,19,22,24). It has been proposed that overproduced rCGG repeats in FXTAS sequester specific RNA-binding proteins and reduce their ability to perform their normal cellular functions, thereby contributing significantly to the pathology of this disorder (19,21,22,24–28).

The disease initially identified as a neurodegenerative disorder was complicated with a broad spectrum of clinical abnormalities among PM carriers, which are apparent both at an early age and most likely in a larger cross-section of carriers than those who develop FXTAS later in life (29,30). The PM carrying children were found to have elevated levels of FMR1 mRNA and comorbid neurodevelopmental disorders such as autism and attention-deficit/hyperactivity disorder (ADHD) (30–32). A study of 43 male subjects including 27 PM carriers showed a significant co-occurrence of autism spectrum disorder (ASD) and ADHD in both proband and non-proband PM carriers (33). Chonchaiya et al. (34) also observed an increased occurrence of autism and ASD features in PM carriers with respect to sibling control. A national survey of families with FXS children, which included 256 PM subjects, revealed increased cases of developmental abnormalities in PM carriers (35).

Experimental investigations focused on early abnormalities have shown mitochondrial dysfunction in fibroblasts and brain samples of PM carriers with and without FXTAS (36,37). Embryonic fibroblasts from the KI mouse have shown abnormal lamin A/C architecture with loss of ring-like nuclear staining (38). Abnormalities during embryonic development include neuronal migration defects, reduced dendrite length and dendritic complexity (39). Corroborating studies from the second KI mouse have shown dendritic morphology abnormalities in cultured hippocampal neurons (40–42). PM KI mice also showed behavioral and cognitive deficits. In behavioral studies, there were progressive deficits in spatial processing (but no motor involvement) in mice as young as 12 weeks (42,43). In addition, female PM carriers aged between 21 and 42 years showed early neurocognitive impairments compared to motor/cognitive impairments associated with FXTAS (44).

The Drosophila disease model of FXTAS has provided experimental evidence for RNA-mediated pathogenic mechanisms. This model has generally been employed for genetic modifier screens because of its convenient phenotypic readouts in the eyes. Although several important proteins/pathways have been identified through this system (19,21,22,26,27,45,46), the eye phenotype, however, represents a late stage when the neurons have deteriorated. Larval neuromuscular junctions (NMJs) are large synapses that are easily accessible for investigation. They offer leeway for a detailed temporal and functional analysis, making extrapolations more accurate (47,48). Hence, we sought to explore the early pathological consequences of fragile X PM rCGG repeats on NMJ structure and function using the Drosophila disease model of FXTAS.

We show that fragile X PM rCGG repeats impair larval locomotion and cause a significant change in the overall morphology of synapses at the larval neuromuscular junction. The total synapse span, including the number of branches and boutons was significantly reduced compared to controls. However, the size of boutons, particularly at the terminals was larger than normal. Aberrant bouton morphology is associated with a significant decrease in the density of Bruchpilot (Brp) protein in active zones, the sites of vesicular release. In accord, rCGG repeats compromise synaptic transmission. The Brp density defects in the active zones were alleviated by the loss of Basket (bsk), a stress-activated c-Jun N-terminal kinase (JNK) and Drep-2, a novel regulatory synaptic protein belonging to the CIDE-N family of proteins. Together, we show that the expression of fragile X PM rCGG repeats impairs synaptic growth and transmission when expressed in the presynaptic compartment of synapses during larval NMJ development.

Results

Neuronal expression of fragile X premutation riboCGG repeats induce defects in larval locomotion

FXTAS has long been considered a late-onset neurodegenerative disease (49). However, recent studies suggest that FXTAS could be the end stage of a life-long process of neuronal deregulation (30). On this premise, we sought to examine early neuronal phenotypes in transgenic flies with both moderate (pUAST-(CGG)₆₀-EGFP) or strong (pUAST-(CGG)₉₀-EGFP) transgene expression (18,19,21,26,27,50). We evaluated transgene expression levels through RT-PCR and by directing transgene expression to several tissues. The transgenic line with the pUAST-EGFP vector served as a control.

We first analyzed the effects of fragile X PM rCGG repeats in embryos. We directed transgene expression to all developing cells of the peripheral and central nervous system using the elav-Gal4 driver. At 25°C, strong expression of r(CGG)₉₀ causes lethality primarily during embryonic development before larval formation (18). However, moderate expression using r(CGG)₆₀ repeats produced viable embryos. We labeled these embryos with anti-FasII and anti-BP102 antibodies, which recognize different subsets and subcellular compartments of neurons in the central and peripheral nervous system, respectively. A high percentage of these embryos did not display any obvious or gross axonal morphology defects (97% of the CGG embryos (n = 150)). However, during the larval stage, they exhibited severely compromised locomotion. In accord with OK6-Gal4, a motor neuron-specific driver, expression of r(CGG)₉₀ repeats displayed a characteristic larval tail-flip phenotype where the posterior body segments flip upwards after each peristaltic muscle contraction (Fig. 1A). To determine the specificity of this phenotype, we performed a systematic assay based on the locomotion activity. In this assay, wandering third-instar larvae were allowed to acclimatize on the agarose plates and were monitored over time by measuring the number of grids (1 sq.mm) crossed by individual third-instar within a 60 s time window over a test period of 180 s. As shown in Figure 1B, the locomotion of both r(CGG)₆₀ and r(CGG)₉₀ was significantly reduced compared to controls (control versus (CGG)₉₀; 0.55 ± 0.02 versus 0.13 ± 0.01 mm/s; mean ± standard error of mean (SEM); P < 0.0001). Importantly, compared to controls, we did not observe a significant difference in locomotion when similar repeats were driven with the MHC-Gal4 driver, a muscle-specific driver. This indicates that locomotion impairments could originate more in neurons than in muscle cells.

Figure 1

Expression of PM rCGG repeats in neurons causes locomotion activity and larval NMJ defects. (A) Representative larvae showing the characteristic tail-flip phenotype in rCGG expressing larvae (OK6-Gal4/UAS-(CGG)₉₀-EGFP) at 25°C while the control (OK6-Gal4/UAS-EGFP) does not show this behavior. (B) Histograms comparing crawling speed in the mentioned genotypes. Student’s t-test was used for statistical analysis. ^*** represents P ≤ 0.001. Error bars represent mean ± SEM. (C) Representative confocal images of wandering third instar larval neuromuscular junction muscle 6/7 of elav-Gal4; UAS-EGFP, (D) UAS-(CGG)₆₀-EGFP, (E) elav-Gal4; UAS-(CGG)₆₀-EGFP at 25°C, (F) OK6-Gal4/UAS-EGFP (G) OK6-Gal4, UAS-(CGG)₆₀-EGFP and (H) OK6-Gal4/UAS-(CGG)₉₀-EGFP at 25°C co-labeled with antibodies against CSP in red and horseradish peroxidase (HRP) in green. The scale bar represents 10 μm. Histograms show (I) total NMJ length normalized to the surface area of their corresponding muscle (length/area in μm²) × 10³, (J) number of boutons per unit muscle area (number/area in μm²) × 10³, and (K) bouton size in μm². ^*** represents P ≤ 0.001 and ^* represents P ≤ 0.05, ns means non-significant. Error bars represent mean ± SEM. Student’s t-test was used for one to one analysis and One-way ANOVA with Tukey’s post-hoc was used for multiple comparisons.

Neuronal expression of fragile X premutation riboCGG repeats induce defects in the morphology of larval neuromuscular junctions

The fact that neuronal expression of rCGG repeats leads to compromised locomotion in larvae prompted us to examine their neuromuscular morphology. We performed open-book preparations of larval body walls (47,51) and used anti-horseradish peroxidase (HRP) to label the neuronal membrane of motor axons. In larvae expressing moderate (CGG)₆₀-EGFP transgene with the elav-Gal4 driver, motor axons emanating from ventral nerve cord (VNC) followed their normal stereotyped pattern of innervations, however, synaptic arbors appeared altered in their gross morphology. Based on this finding, we systematically examined NMJs for type-Ib and type-Is synaptic terminals on muscle 6/7 of the A2 hemisegment for their characteristic synaptic patterning. To normalize any variations in our observations, we considered several measures. Healthy and viable larvae were selected for analysis. Second, computer-assisted software (52) was used to measure the length of the HRP-positive synaptic arbors. Third, each NMJ measurement was normalized to the surface area of the corresponding muscle. Finally, sufficient samples and controls were included to maintain statistical stringency. As shown in Figure 1C–E, and quantified in Figure 1I–K, the normalized total length of elav-Gal4 driven (CGG)₆₀-EGFP motor neuron terminals showed a modest reduction (22%) with respect to controls, 6.747 ± 0.212 versus 5.241 ± 0.247; mean ± SEM; (Total NMJ length/Area in μm²) × 10³ μm; P < 0.05). The decrease in the synaptic span was accompanied by a 36% escalation in the size of presynaptic varicosities (9.34 ± 0.50 versus 12.7 ± 0.41 μm²; mean ± SEM; P < 0.05). Interestingly, however, restricted expression of (CGG)₉₀-EGFP in motor neurons using the OK6-Gal4 driver led to more severe consequences (Fig. 1F–H; quantified in Fig. 1I–K). In these larvae, the normalized total NMJ length decreased by 30% with respect to controls, 7.25 ± 0.24 versus 5.07 ± 0.18; mean ± SEM (Total NMJ length/area in μm²) × 10³ μm; P < 0.0001), normalized bouton number was decreased by approximately 41% (1.29 ± 0.06 versus 0.76 ± 0.03 × 10³; mean ± SEM; (Bouton number/muscle area in μm²) × 10³; P < 0.0001) and size of presynaptic varicosities escalated nearly two times (10.23 ± 0.78 versus 19.66 ± 1.04 μm²; mean ± SEM; P < 0.0001). On the contrary, expression of EGFP or CGG repeats alone did not produce any of these phenotypic effects (Fig. 1C, D, F; quantified in Fig. 1I–K). To check the effect of rCGG repeats on the post-synapse, we employed various muscle-specific Gal4 lines. Stronger Mef-2 Gal4 led to lethality at the first instar stages, while MHC-Gal4 induced lethality at the pupal stages. However, compared to controls, the total NMJ length, bouton number and size of presynaptic varicosities appeared unaffected (quantified in Fig. 1I–K). Taken together, we demonstrated that fragile X rCGG repeats impair larval NMJ morphology specifically when expressed in neurons.

Neuronal expression of fragile X premutation riboCGG repeats affects the remodeling of synapses during larval development

Because boutons are continuously being added or selectively eliminated during their inception from embryonic to late third instar larval development, we asked at what stage rCGG repeat expression specifically manifests the NMJ phenotype. The UAS/GAL4 binary system allowed us to conditionally modulate transgene expression levels by shifting fly cultures to different temperatures (18). We performed crosses of UAS-(CGG)₉₀-EGFP/Cyo with the OK6-Gal4 driver at 18°C and continued this temperature until wandering larvae carrying the genotype (OK6-Gal4/(CGG)₉₀-EGFP) were ready for dissection. No abnormalities in average synaptic growth and bouton number were identified in them with respect to controls; 1.29 ± 0.03 versus 1.26 ± 0.04; mean ± SEM; (Bouton number/muscle area in μm²) × 10³. However, when the same progenies (OK6-Gal4/(CGG)₉₀-EGFP) were shifted to 25°C post embryogenesis to increase transgene expression level, shorter synapses with enlarged boutons similar to the phenotypes shown in Figure 1H were observed; 1.30 ± 0.05 versus 0.75 ± 0.03; mean ± SEM; (Bouton number/Muscle Area in μm²) × 10³; P < 0.0001) (Fig. 2A, quantification). In contrast, the same crosses acclimatized to lay embryos at 25°C, and progenies shifted to 18°C post-embryogenesis, did not produce any NMJ phenotype in third-instar larvae. These results suggest that the phenotype observed upon fragile X PM rCGG repeats commences during larval development. Importantly, under similar treatments, flies expressing EGFP only did not show such abnormalities.

Figure 2

PM rCGG repeats in neurons affects synapse remodeling. (A) Histograms show the number of boutons per unit muscle area in OK6-Gal4/UAS-EGFP, and OK6-Gal4/UAS-(CGG)₉₀-EGFP at 18°C and 25°C. ^*** represents P ≤ 0.001. (B) Histograms show average fluorescence intensity of various glutamate receptor subunits (GluRIIA; GluRIIB; GluRIIC), SSR (Dlg), CSP and Syt normalized to either HRP or Dlg taken from 4 boutons per NMJ as indicated in OK6-Gal4/UAS-EGFP and OK6-Gal4/UAS-(CGG)₉₀-EGFP at 25°C. Students t-test was used for analysis. Representative confocal images showing type Ib boutons of muscle 4 in (C) OK6-Gal4/UAS-EGFP versus OK6-Gal4/UAS-(CGG)₉₀-EGFP and (D) UAS-EGFP/+; MHC-Gal4/+ versus UAS-(CGG)₉₀-EGFP/+; MHC-Gal4/+ at 25°C immunolabeled with CSP or Syt shown in red, and HRP in green. Note the uneven labeling of CSP and Syt. Scale bar represents 5 μm in each case.

Many transient structures have been used to characterize the stages of synapse development. These include ‘ghost boutons,’ presynaptic retractions (footprints) and satellite boutons (51). These structures are characterized and differentiated based on the presence or absence of specific molecular markers (51). We performed co-immunolabeling on NMJ fillets to simultaneously reveal both the pre- and post-synaptic elements of synapses. Most of the boutons on muscle 4 or 6 of elav-Gal4 driven (CGG)₆₀-EGFP repeats were marked by simultaneous retention of both postsynaptic (Dlg or glutamate receptors) and presynaptic (HRP, synaptotagmin or cysteine string protein [CSP]) structures. Therefore, no signs of postsynaptic densities lacking presynaptic components or synaptic retractions were observed. In accordance, normalization of fluorescence intensities from the same bouton showed the intactness of synapses (Fig. 2B, quantification). To check the specificity of this phenotype, we used a strong expression line of (CGG)₉₀-EGFP using an OK6-Gal4 driver. Interestingly, however, presynaptic expression of (CGG)₉₀-EGFP showed a high percentage of boutons with aberrant and uneven localization of vesicle markers, including CSP and synaptotagmin 1 (Fig. 2C). In normal (OK6-Gal4/UAS-EGFP), both CSP and synaptotagmin showed a ring-like localization along the periphery of the boutons (Fig. 2C). Such boutons retained discs large (Dlg) and HRP labeling, and therefore did not show signs of retraction (Fig. 3A and B). No such abnormalities were detected when the expression was driven with the MHC-Gal4 driver (Fig. 2D). This corroborates our observation that fragile X PM rCGG repeats has severe consequences when driven neuronally.

Figure 3

Expression of PM rCGG repeats decreases the number of presynaptic active zones. (A, B) Representative images showing NMJs of muscle 6/7 of (A) OK6-Gal4/UAS-EGFP and (B) OK6-Gal4/UAS-(CGG)₉₀-EGFP. Scale bar represents 10 μm. (C–E) Representative images showing type Ib boutons from NMJs of muscle 4 of (C) OK6-Gal4/UAS-EGFP and OK6-Gal4/UAS-(CGG)₉₀-EGFP (D) elav-Gal4; UAS-EGFP and elav-Gal4; UAS-(CGG)₆₀-EGFP at 25°C, and (E) elav-Gal4/+; UAS-EGFP/+ and elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+ at 18°C. Brp is shown in red and HRP in green. Scale bar represents 5 μm. (F) Histogram plotting of Brp puncta per bouton of indicated genotypes. Error bars represent SEM, ^*** represents P ≤ 0.001. Student’s t-test was used for statistical analysis.

Neuronal expression of fragile X premutation riboCGG repeats reduces the density of Bruchpilot protein at synapses

Boutons contain multiple active zones (neurotransmitter release sites). Therefore, we asked whether altered vesicle markers or an increase in bouton size were associated with the corresponding changes in the active zones. Because OK6-Gal4 driven (CGG)₉₀-EGFP transgene produces prominent defects, we labeled them with an active zone-specific antibody NC82 that recognizes the Bruchpilot protein (Brp). Brp is a key component of the presynaptic active zone required for active zone assembly and vesicle release (53–55). As shown in Figure 3C and quantified in Figure 3F, whole synapse analysis of muscle 4 clearly shows that r(CGG)₉₀ led to a remarkable reduction in the density of Brp. However, this decrease in Brp could be a consequence of the strong morphological defects associated with them. Therefore, we sought to analyze such defects under moderate rCGG repeat expressions. This was achieved by using either the (CGG)₆₀-EGFP transgene (Fig. 3D, quantified in Fig. 3F) or the growing transgene carrying a single copy of (CGG)₉₀-EGFP repeats at 18°C (Fig. 3E). Interestingly, in both cases, a significant reduction in the density of the average number of Brp punctae was observed while maintaining morphological integrity. The reduction in Brp density seemed to be specific to NMJ as analysis of larval segmental nerves and VNC showed uniform level and distribution of Brp punctae with no signs of aggregation (Fig. 4A and B, quantified in Fig. 4C). Also, in comparison to controls, the total amount of Brp in lysates prepared from the adult brains expressing CGG repeats remained unchanged (Fig. 4D). Because active zones are normally apposed to a glutamate receptor (GluR) cluster (51), we asked whether the reduction in Brp punctae has some repercussions on them. Therefore, we analyzed the levels of DGluRIIC, a common subunit of ionotropic glutamate receptors at the NMJ. Although most of the Brp punctae appeared normally opposed by glutamate receptors, some of the Brp punctae did not precede DGluRIIC labeling (Fig. 4E). These results corroborate our above result, which implies that active zone formation proceeds normally, while subsequent synaptic remodeling events may cause a significant drop in their density. No such phenotype was observed in any transgenic line when the expression was driven with the MHC-Gal4 driver.

Figure 4

Expression of PM rCGG repeats compromises pairing of glutamate receptors to presynaptic active zones. (A) Confocal images of larval third instar ventral nerve cords of elav-Gal4/+; UAS-EGFP/+ and elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+ at 18°C. Red represents Brp. Scale bar represents 20 μm. (B) Images showing Brp puncta in red in the segmental nerves of third instar larvae shown in green of elav-Gal4/+; UAS-EGFP/+ and elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+ at 18°C. Scale bar represents 20 μm. (C) Fluorescence intensity of Brp normalized to α-adaptin in the ventral nerve cord of the mentioned genotypes. Error bars represent mean ± SEM. Student’s t-test was used for analysis. (D) Western blot showing Brp protein levels in the adult brains of elav-Gal4/+; UAS-EGFP/+ and elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+ flies at 18°C. (E) Representative images showing type Ib boutons of muscle 4 overexpressing rCGG repeats immunolabeled with GluRIIC shown in green and Brp in red of elav-Gal4/+; UAS-EGFP/+ and elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+ at 18°C. Note the absence of active zones opposite to many glutamate receptor fields. Scale bar represents 3 μm.

Fragile X premutation riboCGG repeats induce defects in synaptic transmission that correlate with the alteration in synapse morphology

Structural changes in synapses are often accompanied by altered synaptic function. Therefore, we examined the consequences of CGG expression on synaptic transmission at the NMJs (56). We performed intracellular electrophysiological analyses on muscle 6/7 of the A2 hemisegment of the third instar larval body walls. Both moderate (CGG)₆₀ and strong (CGG)₉₀ repeats (at 18°C) in comparison to controls exhibited a significant reduction in nerve transmission upon the arrival of impulses, as was evident by a decrease in evoked excitatory junction potentials (EJP) amplitude (Fig. 5A–F). EJP amplitude was reduced by 40% and 46% in r(CGG)₉₀/+ and homozygous r(CGG)₆₀ expressing larvae, respectively. Reduction in EJP amplitude is often a result of reduced quantal size (post-synaptic response to single synaptic vesicle fusions) represented by miniature excitatory junction potential (mEJP) amplitude, quantal content (total synaptic vesicles fused per EJP) or a glitch in both. Assessment of mEJPs showed no reduction in r(CGG)₉₀/+ expressing larvae, excluding the role of quantal size in defective synaptic transmission. However, quantal content was reduced by 53% in r(CGG)₉₀/+ and 46% in r(CGG)₆₀ expressing larvae, indicating that synaptic transmission is perturbed by a decrease in the number of vesicles fusing per EJP (Fig. 5J). This is consistent with our earlier observations, where we found that the number of vesicle-releasing sites was severely reduced upon expression of PM rCGG repeats. Interestingly, we observed a significant surge in the mEJP frequency with a nearly 40% increase in r(CGG)₉₀/+ and 50% in r(CGG)₆₀ expressing larvae. The mEJPs are representative of the basal synaptic transmission indicative of the feedback mechanism, which functions when the post-synapse receives less input from the presynapse upon rCGG repeat expression. As aforementioned, glutamate receptor levels were unchanged upon rCGG expression, consistent with the unrepressed quantal size, despite severely impaired release machinery. This endorses that receptor clusters are normal and synaptic perturbations caused by rCGG repeats are not mediated through these post-synaptic components. Therefore, electrophysiological observations show that reduced release sites lead to a decrease in the release of synaptic vesicles, causing altered synaptic transmission.

Figure 5

Synaptic transmission is severely affected by the expression of PM CGG repeats. (A–F) Representative evoked junction potential and miniature junction potential traces of (A) elav-Gal4/+; UAS-EGFP/+ (B) UAS-(CGG)₉₀-EGFP/+ and (C) elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+ at 18°C. (D) elav-Gal4; UAS-EGFP. (E) UAS-(CGG)₆₀-EGFP (F) elav-Gal4;UAS-(CGG)₆₀-EGFP at 25°C. (G–J) Histograms show (G) mEJP amplitude, (H) mEJP frequency, (I) EJP amplitude and (J) quantal content in the indicated genotypes. Error bars represent SEM. ^* represents P ≤ 0.05, ^*** represents P ≤ 0.001. One-way ANOVA followed by post hoc Tukey’s multiple comparison test was used for statistical analysis.

Fragile X premutation riboCGG repeats induced Brp active zone defects are alleviated from the loss of basket and Drep-2 proteins

The glutamate receptors in rCGG repeat expressing flies sporadically lose their respective Brp punctae in the active zones. Since fragile X PM rCGG repeats are known to induce progressive degeneration in neurons (18), we sought to test pathways associated with chronic neurodegeneration in mediating larval NMJ defects identified in this study. We used anti-Dcp1 and puc-lacZ reporter, a well-established functional readout of JNK activation (57), and found that levels of Death caspase-1 (Dcp-1) and Puckered (Puc) were upregulated in the VNC of larvae expressing r(CGG)₉₀ repeats (Fig. 6A–C). Dcp-1 is a member of the caspase family of ICE/CED-3 proteases and is thought to play a role in apoptosis or programmed cell death (58). Puc is a JNK-specific phosphatase expressed following JNK activation and provides a negative feedback loop to the JNK pathway (57). Since these pathways are reported to play a role in the development and degeneration of neurons (59,60), we investigated the genetic interaction between rCGG repeats and selective genes involved in these pathways. We identified Basket (Bsk), a fly orthologue of human JNK-1, -2, -3 and Drep-2, a member of the CIDE-N domain proteins that can mitigate Brp loss in the active zones of synapses expressing rCGG repeats. Mutants lacking the Basket (bsk¹) are embryonically lethal. However, bsk¹/+ and Drep-2^ex13 null mutants are adult viable and show normal NMJ and eye facet morphology (61,62). Larval NMJ analysis of these mutants and (CGG)₉₀/bsk¹ or (CGG)₉₀, Drep-2^ex13/Drep-2^ex13 recombinant combinations showed no apparent morphological differences between the genotypes. Moreover, mutations did not restore rCGG-induced lethality or the rough eye phenotype induced by rCGG repeats when driven with a Gmr-GAL4 driver (Gmr-Gal4/UAS-EGFP versus Gmr-Gal4,(CGG)₉₀/+). Concurrently, however, the (CGG)₉₀/bsk¹ or recombinant (CGG)₉₀, Drep-2^ex13/Drep-2^ex13 larvae specifically alleviated the loss of Brp punctae in the active zones (Fig. 6D and E). While these findings do not indicate any substantial correlation with the rCGG repeat sequestration mechanism, the effects of RAN translational products remain to be investigated. Two independent studies have reported that Bsk and Drep-2 signaling interfaces with Brp. Specifically, Bsk was shown to rescue the loss of active zone defects due to axonal injury in the synapses (63). Similarly, Drep-2 has been shown to be important for the regulation of synaptic signaling and plasticity through its association with Brp and DmGluRA (62). Therefore, both Bsk and Drep-2 are perfectly positioned to link a variety of environmental conditions with Brp in the active zones. On this basis, we performed a selective interaction screen between rCGG repeats and the pathways corresponding to the effectors and stimulators of the Bsk pathway (Fig. 7A). However, despite our efforts, specific pathways that could mediate these phenotypes have remained elusive in our study. While this could argue for a context or a dosage-dependent interaction, impaired regulation of Bsk, Drep-2 or caspases due to chronic effects of rCGG repeats could mediate neuronal defects.

Figure 6

Loss of Bsk and Drep-2 mitigates the PM rCGG repeat induced loss of active zones. (A) Representative confocal images showing Dcp-1 protein levels in third instar larval ventral nerve cords of elav-Gal4/+; UAS-EGFP/+ and elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+. (B) Confocal images showing Puc protein levels in third instar larval ventral nerve cords of elav-Gal4/+;UAS-EGFP/+ and elav-Gal4/+;UAS-(CGG)₉₀-EGFP/+. Scale bar represents 20 μm. (C) Histogram representing Dcp-1 and Puc levels in elav-Gal4/+; UAS-EGFP/+ versus elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+. Comparisons made with Students t-test. ^** represents P ≤ 0.01. Error bars represent mean ± SEM. (D) Confocal images showing the number of active zones in (1) elav-Gal4/+; UAS-EGFP/+ (2) elav-Gal4/+; UAS-(CGG)₉₀-EGFP/+ (3) Drep-2^ex13/Drep-2^ex13 (4) elav-Gal4/+; UAS-(CGG)₉₀-EGFP, Drep-2^ex13/Drep-2^ex13 (5) bsk¹/+. (6) elav-Gal4/+; UAS-(CGG)₉₀-EGFP/bsk¹ larvae. Scale bar represents 5 μm. (E) Histograms plotting Brp puncta per bouton in the designated genotypes. ^** represents P ≤ 0.01, ^*** represents P ≤ 0.001 and error bars represent mean ± SEM. Comparisons made with one-way ANOVA followed by post hoc Tukey’s multiple comparison test.

Figure 7

Misregulation of Bsk or Drep-2 involved in synapse development and plasticity could lead to the neuronal defects associated with PM rCGG repeats. (A) Proposed model of Drep-2 or Bsk modulating fragile X PM rCGG repeat-mediated NMJ active zone defects. rCGG repeats could interfere with the known Drep-2 or Bsk pathway. rCGG repeats through RAN translation products, FMRpolyG could impair UPS, which in turn impairs the regulation of Bsk pathway. Mutation in Bsk or Drep-2 could suppress rCGG repeat Brp active zone defects. Mutant alleles tested for genetic interactions are shown in parenthesis. Shown are observed interactions (solid line), failed interactions (broken lines), potential interaction (double arrowed lines), interaction found through immunoprecipitation (double lines). (B) Model representing the overall NMJ phenotype in various genetic backgrounds. rCGG repeats affect the overall NMJ morphology. Total NMJ length and bouton number are reduced. Boutons are increased in size with reduced density of Brp at active zones.

Discussion

FXTAS is a late-onset neurodegenerative disorder that is distinct from neurodevelopmental fragile X syndrome caused by FMR1 gene silencing (64,65). The neurological phenotypes in FXTAS patients, likely stemming from RNA toxicity and repeat-associated non-AUG (RAN) mediated protein toxicity, usually appear after the fifth decade of life (18,19,46,66). However, increasing clinical data have broadened the pathological foray of FXTAS. Several neurodevelopmental and psychological deficits have been observed in FXTAS patients. Thus, FXTAS is no longer recognized as pure neurodegenerative disorder (29,30,33,35,67).

How expanded rCGG repeat-mediated pathogenesis embarks on neurodevelopmental processes and causes misregulation is not well understood. Hence, it is imperative to trace the early pathological abnormalities caused by the fragile X PM rCGG repeats. Here, using a transgenic Drosophila model, we spatiotemporally tracked the early events of the rCGG-mediated pathogenesis in neurons. We show that the early assaults led by the expanded CGG repeats have a deleterious influence on the structure and function of synapses in vivo (Fig. 7B). Analysis of NMJ parameters showed reduced synaptic span and larger boutons than controls (Fig. 1C–H, I–K, quantification), indicating that expanded repeats can potentially cause cell-autonomous defects in motor neurons. Sub-synaptic analysis also revealed defects in the underlying architecture with a reduced density of presynaptic Brp (Fig. 7B). Brp has sequence homology to ELKS/CAST and is a key component of presynaptic active zones (neurotransmitter release sites) (54). In flies, disruption of its action alters active zones and may leave GluRs orphan without any preceding active zones (53). Brp normally apposes their GluR receptor fields in the rCGG synaptic boutons. However, some synapses lose Brp punctae in the active zones to their respective DGluRIIC labeling (Fig. 4E). This occurs sporadically and not in distal synapses as attributed to neurodegenerative processes of synaptic disassembly or active zone decline (68). However, presynaptic membranes are homogenous with a normal surround of subsynaptic reticulum (SSR). In addition, Brp is normally expressed and delivered to synapses. Significantly, this manifests as a strong reduction in the quantal content, a measure of total synaptic vesicles released per excitation potential (Fig. 5J). This endorses that receptor clusters are normal, and synapse perturbation caused by rCGG repeats is not mediated through these postsynaptic compartments. This is evidently due to the reduction in active zones known to modulate synaptic communication (53). Together, this implies that synapses are intact and active zone formation proceeds normally.

How might fragile X PM rCGG repeats mediate its effect on synapses? We noticed that Basket (Bsk) and Drep-2 loss specifically mitigated Brp active zone defects in larvae expressing rCGG repeats. The Bsk is a component of the stress-activated JNK pathway. Its complexity is perplexing, with multiple positive and negative regulators underlying the diversity of functions. It is reported to promote synapse growth through its downstream transcription effectors, D-Fos and D-Jun (69,70). Bsk also rescues the active zone depreciation and synapse overgrowth in unc-104 mutants via the axonal damage signaling kinase protein Wallenda, an upstream signaling partner of Bsk (63). Drep-2 belongs to the CIDE-N domain containing DNA fragmentation factor (Dff) family of proteins and is assumed to be involved in the apoptotic regulation of DNA degradation (62,71). It is expressed in the central synapses and is considered to be important for the regulation of synaptic signaling and plasticity through its association with Brp and DmGluRA (62). Although the hierarchy of regulatory mechanisms between Brp and Bsk or Drep-2 remains elusive, in their canonical pathways, both Bsk and Drep-2 are uniquely positioned to integrate diverse pathways in a context-dependent manner with synaptic growth and plasticity (Fig. 7A). Because the rCGG-induced active zone phenotypes were specifically mitigated by the loss of Bsk or Drep-2 proteins, the sequestration-based model does not appear to have a significant correlation in this case. Repeat-associated non-AUG (RAN) translation of rCGG repeats leads to the production of toxic homo-polypeptides, such as FMRpolyG, which induces impairments in the ubiquitin–proteasome system (UPS) (72). The UPS in a context-specific manner anticipates its role in the regulation of the JNK pathway (69,73,74). Therefore, (chronic) RAN translation products could potentially implicate their effects on the JNK pathway. A more direct approach to determining the contribution of RAN translation to observed phenotypic manifestations would be to investigate these phenotypes in specific transgenic flies carrying CGG repeats that lack RAN translation or enhance FMRpolyG production.

As both the JNK pathway and caspases per se appeared to be up-regulated in the VNC of larvae expressing r(CGG)₉₀ repeats, therefore, at first glance, it might appear surprising that other proteins corresponding to these pathways, including autophagy, and caspases had modest effects on the amelioration of rCGG induced synapse defects. However, interactions with rCGG repeats can act in a context or dosage-dependent manner. Concurrent to this statement, Drep-2 has been previously implicated in the suppression of rCGG repeats-induced eye degeneration (26). This is in contrast to our observation, where Drep-2 loss appears to compensate for Brp active zone defects at larval NMJs. In larval NMJs, there are no reports of structural defects due to Drep-2 loss, rather, Drep-2 is thought to be important for learning and memory (62). Nevertheless, based on our findings, fragile X PM rCGG affecting multiple and diverse pathways at synapses is a possible scenario, which is now open for further investigation.

Synaptic defects are common in early neurological diseases, including FXS, ASD, Schizophrenia and Bipolar disorder (30,75–78). On the other hand, in late-onset neurodegeneration, such as Alzheimer’s (ad), Parkinson’s (PD) and Huntington’s (HD) diseases, cell death is the inevitable end-result of an ongoing pathophysiological cascade. These diseases also converge to the same structural and pathological features, such as mitochondrial abnormalities (79,80) and synaptic defects (81–83), and in some cases cause the absence of an assortment of presynaptic proteins such as active zone components and vesicle-associated proteins (81–85). PM repeats cause early defects in neuronal morphology (39,40), spontaneous calcium oscillations and clustered burst firing in animal models (41). Solidifying these observations is the growing clinical data showing alterations in the brain well before the onset of FXTAS. These include gray matter changes (86), white matter tract changes (87) and functional MRI recordings (88). Thus, it is evident that comorbid neurodevelopmental conditions with FXTAS are growing; however, the relationship between their manifestations remains obscure. As we have tried to establish the role of fragile X PM in mediating neurodevelopmental defects, in the case of fragile X syndrome, the same might hold for the longer repeats in the full mutation range before methylation of the promoter. Although flies carrying longer (CGG)_>200 repeats are unstable, our findings suggest that rCGG repeats cause synaptic abnormalities. Whether these abnormalities proceed to neuronal cell death, as seen in FXTAS, remain to be investigated. However, these abnormalities could provide a better understanding of the underlying shared mechanisms among comorbid neurological disorders. Such information will lead to improved disease biomarkers and novel therapeutic approaches.

In summary, our observation in a transgenic model of fragile X PM rCGG repeats shows that the structural and functional alterations in synapses caused by fragile X PM rCGG repeats may contribute to FXTAS pathogenesis, and the mechanism could be via JNK/Bsk and Drep-2 because their loss can restore Brp density defects at active zones seen upon fragile X PM rCGG expression.

Materials and Methods

Fly stocks and genetics

All fly strains were reared on standard cornmeal media at 25°C, unless otherwise mentioned. Transgenic flies with both moderate (pUAST-(CGG)₆₀-EGFP) or strong (pUAST-(CGG)₉₀-EGFP) transgene expression and control (pUAST-EGFP) used in this study were generated previously (18). The presence of CGG repeats in the transgenic flies was confirmed by polymerase chain reaction (PCR) using C and F primers, as described previously (18,89). The expression of these repeats was checked by reverse transcription PCR using primers against the eGFP tag (90). Stocks elav-Gal4, OK6-Gal4, bsk¹ and Df(3 L)H99 were procured from the Bloomington Drosophila Stock Center, Indiana University. The Drep-2^ex13 fly line was obtained from the laboratory of Stephen Sigrist (62); puc-lacZ (puc^E69) (57), wnd¹ (73). The Atg¹ (91) was obtained from the laboratory of Vimlesh Kumar.

Immunohistochemistry and antibodies

For NMJ immunohistochemistry wandering third instar larvae were dissected in ice-cold calcium-free hemolymph-like saline (HL3) solution and fixed in 4% paraformaldehyde in 1× phosphate-buffered saline (PBS), pH 7.2 for 30 min, or in Bouin’s fixative (glutamate receptor staining) for 5 min. After washing with PBS containing 0.2% Triton X-100 (PBT), larval fillets were blocked in 5% bovine serum albumin (BSA) in PBT for 1 h and then incubated overnight with the primary antibody in PBT at 4°C. The larvae were then washed in PBT and incubated in secondary antibody in PBT for 1 h, washed and mounted in VectaShield (Vector Laboratories, Burlingame, CA) on slides. The following monoclonal antibodies from Developmental Studies Hybridoma Bank (University of Iowa, USA) were used for immunostaining: anti-CSP (1:50), anti-Brp^Nc82 (1:50), anti-β-Gal (1:50), and anti-GluRIIA (1:50). Rabbit anti-GluRIIB (1:1000) and anti-GluRIIC (1:5000) were provided by Aaron DiAntonio (92). Rabbit anti-α-adaptin (93) was used as an internal control for Brp staining in larval brains. Rabbit anti-cleaved Drosophila Death caspase-1 (1:200, Cell Signaling Technology #9578) and rabbit anti-Drep-2 (1:500) (62). Secondary antibodies conjugated to Alexa Fluor 488, Alexa Fluor 568, Alexa Fluor 555, and Alexa Fluor 633 (Molecular Probes, Thermo Fischer Scientific) were used at 1:800 dilution. Anti-HRP antibodies conjugated to Alexa-488 or Rhodamine were used at a 1:1000 dilution.

Larval crawling assay

The third instar larvae were gently collected and washed with deionized water. The larvae were subsequently transferred onto a 2% agarose gel in a transparent Petri dish placed on top of a gridline-marked surface. The larvae were allowed to acclimatize for some time before quantification. The average distance traversed by larvae in 60 s was calculated for each genotype. The experiments were performed under uncrowded conditions. Statistical analyses based on Students t-test were performed using GraphPad Prism software (GraphPad Software, San Diego).

Electrophysiology

The Intracellular electrophysiological recordings were performed at room temperature. Glass microelectrodes filled with 3 M KCl with resistance between 12 and 20 Ω were used. Experiments were performed on third instar larval NMJs (muscle 6, A2 hemisegment) in HL3 saline with the following composition (in mm): NaCl 70, KCl 5, MgCl₂ 20, NaHCO₃ 10, sucrose 115, trehalose 5 and HEPES 5 with pH 7.2 supplemented with 1.5 mm calcium. Recordings were performed from muscles with resting potential between −60 and −75 mV and input resistance of ≥4 MΩ. For recording EJPs, nerves were stimulated at 1 Hz and spontaneous release events lasted for 60 s. The quantal content was estimated by the ratio of the average EJP to the average mEJP amplitude for each NMJ. The signals were amplified using Axoclamp 900A and data acquisition was performed using Digidata 1440A and pClamp10 software (Axon Instruments, Molecular Devices, USA). Mini-Analysis (Synaptosoft) was used for the data analysis.

Western blotting

Third instar larval brains were manually homogenized in 1.5 ml Eppendorf tubes at 65°C using a micropestle in an equal amount of 2× Laemmli buffer and the samples were heated at 95°C for 10 min. The debris was pelleted by centrifugation at 13 000 × g for 5 min and the protein sample equivalent to five heads was resolved on 8% denaturing SDS-PAGE. Proteins were transferred to a Hybond-PVDF-LFP membrane (Amersham, GE Healthcare Life Sciences) and blocked with 5% skimmed milk in 1× Tris-buffered saline, supplemented with 0.1% Tween-20 (TBST) for 1 h at room temperature. The membrane was probed with primary antibodies against Brp^Nc82 (1:500), rabbit anti-Drep-2 (1:5000) (62) and α-Actin (1:5000, Cell Signaling) in 1× TBST containing 2% BSA at 4°C overnight followed by washing steps in TBST. This was followed by incubation with secondary antibodies conjugated to HRP in 1× TBST (1:10 000) for 1 h followed by washing steps. The membranes were incubated with ECL Prime western blotting Detection Reagent (GE Healthcare) and signals were acquired using a Fuji LAS-4000 Image Analyzer (Fuji Film, Tokyo, Japan). The images were analyzed using Image Gauge software (Fuji Film).

Image acquisition and analysis

Images were captured using a Zeiss LSM780 confocal microscope. NMJ morphometric analyses were performed at muscle 6/7 of abdominal segment 2, except for Brp puncta quantification, which was performed at muscle 4 of abdominal segment 2 for convenience. Neuronal membranes were labeled with anti-HRP antibody and boutons were specifically labeled with vesicle-specific anti-CSP antibody. Images for fluorescence quantification were acquired using the same parameters and settings. Fluorescence intensities were quantified using Image J (National Institute of Health) or Zen software (Carl Zeiss). Data are reported as the mean ± SEM. Images were processed using Adobe Photoshop (Adobe Systems, San Jose, CA, USA). For statistical analysis, a two-tailed Student’s t-test was used for experiments involving two data sets; otherwise, one-way ANOVA with Tukey’s multiple comparison post-test was employed. The asterisks in the figures represent the level of significance: ^*, P ≤ 0.05; ^**, P ≤ 0.01 and ^***, P ≤ 0.001.

Acknowledgements

We are especially grateful to Prof. Peng Jin for providing the tools required for the study (transgenic flies carrying CGG repeats and controls). We are thankful to Prof. Stephen Sigrist, Prof. Aaron DiAntonio, the Developmental Studies Hybridoma Bank, and Bloomington Stock Center for providing antibodies and important fly lines. We are very thankful to Dr Vimlesh Kumar for the Confocal Facility at IISER Bhopal.

Conflict of Interest statement. The author(s) declare that they have no competing interests.

Funding

S.A.B. is a recipient of a research fellowship from the Council of Scientific and Industrial Research, New Delhi (09/251/(0058)/2014-EMR-I). A.Q. was supported by the grant from the Department of Biotechnology, Government of India (BT/RLF/Re-entry/51/2012). A.Q. is a recipient of the Ramalingaswami Re-entry Fellowship and the University Grants Commission-Startup Award. We are thankful to the Department of Biotechnology, Govt of India (BT/RLF/Re-entry/51/2012) and UGC-BSR for funding this project to Abrar Qurashi. We are thankful to the Council of Scientific and Industrial Research, New Delhi (09/251/(0058)/2014-EMR-I) for a research fellowship to Sajad A. Bhat.