Protective role of PASH-1 in CGG repeat-driven RNA and protein toxicity in FXTAS

Summary

Fragile X-associated tremor/ataxia syndrome (FXTAS) is caused by CGG repeat expansions in FMR1, leading to RNA toxicity and toxic FMRpolyG peptide from abnormal translation. Using a Caenorhabditis elegans model, we generated single-copy insertions of the human FMR1 5′ UTR containing 0, 16, or 99 CGG repeats under a pan-neuronal promoter. Worms expressing 99 CGG repeats showed impaired motility, altered neuronal morphology, and disrupted miRNA homeostasis. Co-expression of PASH-1, the C. elegans ortholog of a miRNA-processing DGCR8 sequestered in FXTAS, mitigated both RNA- and peptide-mediated toxicity, restoring locomotion, neuronal structure, and miRNA regulation balance. Removing FMRpolyG improved movement by ∼50%, suggesting RNA toxicity is the primary pathogenesis. Glial 99 CGG expression altered nearby neuronal cilia, disrupting olfaction without affecting movement, revealing non-cell-autonomous toxicity. These findings establish the protective role of PASH-1 against CGG-induced neurotoxicity and underscore C. elegans as a model for dissecting FXTAS mechanisms and exploring therapeutic strategies.

Graphical abstract

Highlights

• •Single copy CGG RNA causes toxicity that disrupts motility and neuronal features
• •PASH-1 lowers toxic CGG RNA levels and restores locomotion and neural integrity
• •Removing FMRpolyG only partly improves movement: primary role of RNA toxicity
• •Glial CGG repeat RNA induces non-cell-autonomous defects in olfactory neuronal cilia

Neuroscience; Cell biology

Introduction

Fragile X-associated tremor/ataxia syndrome (FXTAS) is a progressive neurodegenerative disorder that primarily affects older male carriers of premutation alleles (55–200 CGG repeats) in the 5′ untranslated region (UTR) of the Fragile X Messenger Ribonucleoprotein 1 (FMR1) gene.¹ The core clinical features of FXTAS include intention tremor and cerebellar ataxia, with associated features of parkinsonism, autonomic dysfunction, peripheral neuropathy, and cognitive decline.^2,3,4,5 Neuropathological hallmarks of FXTAS include widespread intranuclear inclusions and white matter disease in multiple brain regions, notably the cerebellum and frontal cortex.^6,7,8,9 Although there are no dominant species among the hundreds of proteins within the inclusions, the more abundant proteins, p62, SUMO2, and ubiquitin, suggest a role for the inclusions in degradation/elimination of damaged/oxidized proteins.^8,9

FXTAS arises from a toxic RNA gain-of-function mechanism believed to be mediated by two major, non-mutually exclusive molecular mechanisms: sequestration of one or more RNA-binding proteins by the expanded-repeat mRNA and/or non-AUG translation through the CGG repeat to create a toxic peptide, designated FMRpolyG. First, premutation carriers exhibit a 2- to 8-fold increase in FMR1 mRNA,^10,11,12 which may sequester specific RNA-binding proteins (RBPs) away from their natural targets, thereby disrupting RNA metabolism.¹³ Expanded CGG repeats have been shown to bind and titrate RBPs such as heterogeneous nuclear ribonucleoprotein (hnRNP) A2/B1 in the cytoplasm, impairing the dendritic delivery of its target BC1 RNA.¹⁴ In Drosophila, the Hrb87F and Hrb98DE homologues of hnRNP A2/B1 were shown to modulate CGG repeat-induced toxicity.¹⁵ Similarly, overexpression of Purα, another RBP that binds CGG repeat RNA, rescues neurodegeneration in flies, although its presence in FXTAS patient inclusions is inconsistent.^13,16,17

Among the RBPs implicated, DiGeorge syndrome critical region 8 (DGCR8) preferentially binds expanded CGG-repeat RNA and appears to bind cooperatively with increasing repeat number, as shown in pull-down experiments using mouse brain nuclei.¹³ DGCR8 is a core component of the microprocessor complex with DROSHA, which cleaves primary miRNAs (pri-miRNAs) into precursor miRNAs (pre-miRNAs) to initiate miRNA biogenesis.¹⁸ In neuronal cells, partial sequestration of this complex into nuclear CGG RNA foci may impair miRNA processing and alter miRNA profiles. Indeed, reduced miRNA expression has been observed in FXTAS patient samples and in Drosophila CGG models.^19,20 Elevated pri-miRNA levels in FXTAS brain tissues further support disrupted processing, and overexpression of DGCR8 partially rescues CGG-induced dendritic defects in neuronal culture.¹³ These findings suggest that RBP sequestration by CGG RNA disrupts mRNA transport, translation, and miRNA biogenesis.

The second proposed model involves repeat-associated non-AUG (RAN) translation, through which expanded CGG repeats are proposed to facilitate the production of toxic polyglycine-containing proteins (FMRpolyG) from a non-canonical start codon (e.g., GUG, CUG, ACG) located upstream of the CGG repeat.^21,22,23 Translation of FMRpolyG is enhanced by elevated FMR1 mRNA. FMRpolyG is detectable in trace amounts by mass spectrometry within the intranuclear inclusions of neural cells⁹ and is also detectable using anti-FMRpolyG antibodies.^23,24 Although there are clear associations between the presence of FMRpolyG, intranuclear inclusions, and neuronal dysfunction,²⁴ any causative relationship is still unclear. Suppression of FMRpolyG in some mouse models ameliorates motor and neuropathological phenotypes,^8,23 whereas other mouse models fail to display any significant behavioral or motor phenotype.²⁵ Thus, studies to date have not clearly established either sequestration or RAN-based models of FXTAS. Indeed, it may be that both mechanisms are operating.

Although CGG-repeat mouse and cell models recapitulate key FXTAS features, including RNA toxicity and protein aggregation, studying neural circuits and non-cell-autonomous mechanisms in these systems remains challenging. The nematode Caenorhabditis elegans offers a genetically tractable alternative with a simple nervous system amenable to comprehensive behavioral analysis. Moreover, the development of numerous promoter constructs has facilitated the engineering of exogenous genes into specific neurons or collections of neurons. Notably, C. elegans lacks both endogenous CGG repeats and an FMR1 ortholog, providing a clean genetic background for studying exogenous CGG repeat toxicity. The phiC31 integrase-mediated recombination technique successfully generated a single-copy insertion of the human FMR1 5′ UTR containing CGG repeats (as described herein), closely mimicking endogenous gene context and expression. With 83% of C. elegans proteins having human homologues²⁶ and the use of tissue-specific promoters, the current experimental model allows precise dissection of cell-autonomous and non-cell-autonomous mechanisms of CGG-induced neurotoxicity.

Here, we demonstrate that expression of PASH-1, the C. elegans ortholog of DGCR8, effectively mitigates RNA and protein toxicities caused by expanded CGG repeats, subsequently restoring impaired motility, neuronal morphology, olfactory function, and miRNA homeostasis. Using a C. elegans model with a precise single-copy insertion of the human FMR1 5′UTR with an expanded CGG repeat sequence, we observed significant age-dependent motor dysfunction that resembled the symptoms of FXTAS. Notably, expression of an RNA that does not permit production of FMRpolyG resulted in approximately 50% reduction in motor dysfunction, directly proportional to lower mRNA levels, supporting the clear RNA gain-of-function in motor pathology. Furthermore, glial-specific expression of CGG repeats caused non-cell-autonomous cognitive impairment, particularly olfactory defects, due to altered ciliary morphology. Our findings underscore the multifaceted cellular toxicity of CGG repeats and identify PASH-1 as a protective modulator and promising therapeutic target for FXTAS.

Results

PASH-1 mitigates CGG RNA overexpression and restores mRNA homeostasis in a C. elegans FXTAS model

To establish a physiologically relevant FXTAS model in C. elegans (which possesses 302 neurons among its 959 somatic cells in the adult stage), we introduced single-copy insertions of the human FMR1 5′ UTR containing 0, 16, or 99 CGG repeats upstream of a GFP reporter, driven by the pan-neuronal unc-119 promoter (Figure 1A). Single-copy insertion via phiC31 integrase-mediated recombination minimized artifacts associated with multicopy arrays and ensured stable, neuron-wide expression.²⁷ The constructs were microinjected into a C. elegans strain BRC0566, which carries a defective unc-119 gene, enabling phenotypic selection of successful integrants based on rescued motility (Figure S1). These strains enable qualitative and quantitative characterization of phenotypic and molecular neuropathological mechanisms mediated by CGG repeats.

Figure 1.

PASH-1 co-expression reverses CGG repeat-induced RNA overexpression in C. elegans

(A) Schematic representation of the constructs used to study CGG repeat expansions in C. elegans. Worm strains include 0CGG (control), 16CGG (normal-repeat control), and 99CGG (clinical FXTAS repeat size), all under the pan-neuronal punc-119 promoter driving GFP expression.

(B) Schematic representation of the construct expressing pash-1 cDNA with a C-terminal HA::mCherry tag, driven by the punc-119 promoter.

(C) Co-expression of PASH-1 restores elevated FMR1 RNA levels in 99CGG worms to near-control levels. RT-qPCR results show that the 99CGG expansion induces an average 6.5-fold increase in FMR1-driven GFP mRNA levels compared with the 16CGG control, indicating RNA overexpression. Co-expression of PASH-1 in 99CGG worms significantly reduces GFP mRNA levels, demonstrating its ability to mitigate RNA overexpression. PCR primer pairs are indicated by red arrows.

(D) mRNA levels of pash-1 in CGG-expanded worms. RT-qPCR analysis showed that pash-1 mRNA levels increased 2.1-fold in PASH-1-overexpressing 99CGG worms, effectively counteracting RNA toxicity. PCR primer pairs are indicated by red arrows. In (C) and (D), each data point represents an independent biological replicate. Data were obtained from six independent experiments, using total RNA isolated on the same day to measure FMR1::GFP and pash-1 mRNA levels. mRNA expression for each genotype was first normalized to that of the housekeeping gene act-3, and fold changes were then calculated relative to the 0CGG strain. The red line indicates no change relative to the 0CGG worms. p-values represent statistical comparisons of fold change between genotypes using one-way ANOVA.

To examine the molecular neuropathology of FXTAS in C. elegans, we assessed the effects of expanded CGG repeats on FMR1 mRNA levels. We performed reverse transcription and quantitative real-time PCR (RT-qPCR) analyses on single-copy insertion strains expressing 0, 16, or 99 CGG repeats. The PCR primers targeted the region spanning the 3′ end of the FMR1 5′ UTR and the GFP coding region, with normalization to the housekeeping gene act-3. RT-qPCR analysis revealed a 6.5-fold increase in FMR1 mRNA levels in 99CGG-expressing worms relative to 0CGG controls (Figure 1C, second bar), recapitulating the elevated mRNA levels observed in FXTAS patients,^10,12 which reflect increased transcriptional activity.¹¹ The 16CGG line exhibited mRNA levels similar to those in 0CGG controls, confirming the threshold repeat number for enhanced transcription. These findings support the utility of the C. elegans model in dissecting CGG repeat-induced pathologies.

Given DGCR8’s established role in CGG RNA binding and miRNA processing, and the identity of PASH-1 as its C. elegans ortholog, we tested whether PASH-1 plays a role in regulating RNA toxicity. PASH-1 was expressed as an HA-tagged mCherry fusion protein under the pan-neuronal unc-119 promoter (Figure 1B). To ensure stable inheritance, the transgenic arrays were integrated through trimethylpsoralen treatment and UV exposure.²⁸ Crossing the integrated strain into 99CGG worms allowed co-expression of both the CGG repeat reporter and PASH-1 in the same neurons. RT-qPCR analyses showed that co-expression of PASH-1 significantly reduced FMR1 mRNA levels in 99CGG worms to near baseline levels (0.9-fold relative to 0CGG; Figure 1C, third bar). Since DGCR8 (the human ortholog of PASH-1) binds only weakly to short, nonpathogenic CGG repeat RNAs,¹³ co-expression of PASH-1 in worms expressing 0CGG or 16CGG did not significantly alter the mRNA levels of these non-toxic repeats (Figure S2).

Quantification of pash-1 mRNA by RT-qPCR analysis showed a 2.1-fold increase in pash-1 mRNA in the 99CGG background (Figure 1D, third bar). Endogenous pash-1 levels were unchanged in 16CGG and 99CGG strains compared with 0CGG (Figure 1D, first and second bars). These results demonstrate that PASH-1 effectively counteracts the overexpression of FMR1 mRNA caused by expanded CGG repeats, likely by restoring mRNA homeostasis. These findings highlight PASH-1 as a critical factor in mitigating the molecular pathologies associated with FXTAS.

Age-dependent crawling ability in FXTAS worm model

FXTAS is a late-onset disorder caused by premutation CGG repeats (55–200 repeats) in the human FMR1 gene, typically manifesting after age 50 years. To investigate age-related motor decline, we evaluated crawling performance in the C. elegans FXTAS model by directly observing larval movement on solid media. L3 larvae were placed at the center of a 1 cm square drawn on a 6 cm nematode growth medium (NGM) plate, and the time taken to exit the square was recorded (Figure 2A). L3 larvae from wild-type worms and worms expressing 0, 16, or 99 CGG repeats showed similar escape times (Figure 2B). Similarly, one-day-old adult worms expressing 0 or 16 CGG repeats exited a 1.5 cm square within 1 min, indicating normal motility (Figure 2C, second and third bars). In contrast, adults with 99 CGG repeats exhibited significant impairments, taking up to 300 s to escape (Figure 2C, fourth bar), indicating motor dysfunction.

Figure 2.

Age-related decline in crawling ability of worms expressing expanded CGG repeats

(A) Schematic of the crawling assay. L3 or adult worms were placed in 1 cm or 1.5 cm squares, respectively, on NGM plates. Under a dissecting microscope, the time each worm required to exit the square was recorded in seconds to assess locomotor ability.

(B–C) Crawling behavior varied with developmental stage and CGG repeat length. In L3 larvae (B), locomotion was comparable across all groups, including wild-type, 0CGG, 16CGG, and 99CGG worms, with each group requiring approximately 47 s to exit the square. Each data point represents the time required for a single worm to exit the square. Data were obtained from one independent experiment, with sample sizes as follows: wild type n = 10, 0CGG n = 13, 16CGG n = 12, 99CGG n = 11. However, in adult worms (C), crawling ability differed significantly depending on CGG repeat length. Wild-type, 0CGG, and 16CGG adults exited the square in approximately 40 s, whereas 99CGG-expressing worms required approximately 292 s. Notably, co-expression of PASH-1 in 99CGG worms significantly improved motility, reducing the exit time to approximately 65 s. Each data point represents the time required for a single worm to exit the square. Data were obtained from one independent experiment, with sample sizes as follows: wild type n = 37, 0CGG n = 33, 16CGG n = 23, 99CGG n = 23, 99CGG+PASH-1 n = 40.

(D) WormTrackAI: Schematic of automated quantification of worm crawling distance using computer vision. Worms were placed on NGM plates and recorded under a fluorescent dissecting microscope. Crawling behavior was videotaped for approximately 30–60 s, and movement was analyzed using centroid-based image tracking.

(E and F) Crawling distances also varied with developmental stage and CGG repeat expression. L3 larvae (E) expressing 0CGG, 16CGG, or 99CGG displayed crawling distances comparable with wild type over 20 s. Each data point represents the crawling distance of a single worm measured over 20 s. Data were obtained from one independent experiment, with five worms analyzed per genotype. In contrast, adult worms (F) expressing 99CGG exhibited significantly reduced crawling distances (average of 2.39 mm) compared with wild-type and 16CGG worms, which traveled 6.82 mm and 6.79 mm on average, respectively. Co-expression of PASH-1 in 99CGG worms partially rescued motility, increasing average travel distance to approximately 5.03 mm. Each data point represents the crawling distance of a single worm measured over 20 s. Data were obtained from one independent experiment, with sample sizes as follows: wild type n = 5, 16CGG n = 5, 99CGG n = 10, 99CGG+PASH-1 n = 12.

(G–J) Representative crawling tracks.

(G and H) show 60-s paths of adult wild-type and 16CGG worms, respectively.

(I and J) show 30-s zoomed images, with (J) highlighting mCherry-labeled adult worms expressing 99CGG with PASH-1. Bars in (B) and (C) represent mean time; bars in (E) and (F) represent mean distance. Error bars represent SEM. Statistical significance was determined by one-way ANOVA. Scale bar indicates 3 mm in (G and H) and 1 mm in (I and J).

To quantify movement, we recorded videos under a fluorescent stereomicroscope (Leica M165 FC) and analyzed worm crawling tracks using WormTrackAI image processing. L3 larvae were recorded for 20 s on NGM plates (Figure 2D), and their crawling distances were quantified via image analysis. Crawling distances of L3 larvae expressing 16 or 99 CGG repeats were comparable with wild-type controls (Figure 2E).

In adult worms, motor performance varied depending on CGG repeat length. Adults expressing 16 CGG repeats exhibited movement comparable with wild-type controls over a 20-s period (Figure 2F, first and second bars; 2G and 2H). In contrast, worms expressing 99 CGG repeats showed a marked reduction in locomotion, approximately 35% of wild-type levels (Figure 2F, third bar; Figure 2I).

To examine whether L3-stage worms expressing 99 CGG repeats, which display near-normal crawling behavior, might express normal levels of FMR1 mRNA, we performed RT-qPCR analysis on synchronized L3 worms expressing 99 CGG repeats (Figure S3). The data indicate that, although these L3 worms exhibit nearly wild-type crawling behavior, their FMR1::GFP mRNA levels are already elevated by approximately 2.3-fold compared with the control (Figure S3, second bar). In contrast, adult 99 CGG worms exhibit an even greater increase (∼4.5-fold; Figure S3, third bar), which correlates with their impaired motility.

In humans, males of any age carrying moderate to large CGG premutations (55–200 repeats) typically show a 2- to 8-fold elevation in FMR1 mRNA levels, although the presence of intranuclear inclusions is not apparent until late adulthood.²⁹ Similarly, Fmr1 premutation mice display early developmental deficits in the absence of inclusions, which form only after 20–40 weeks. For instance, neuronal migration is impaired during embryonic stages,³⁰ and hippocampal neurons cultured from premutation mice show a ∼3.8-fold increase in Fmr1 mRNA levels, along with reduced viability at 21 days in vitro.³¹ Therefore, our findings in C. elegans align well with both human and mouse models of CGG repeat expansion and indicate that cellular dysregulation substantially precedes inclusion formation, indicating that inclusion formation per se is a late event in pathogenesis.

The substantial motor impairment caused by 99 CGG repeats in adult but not larval stages highlights an age-dependent phenotype in the C. elegans FXTAS model. This decline mirrors the progressive nature of FXTAS in humans and provides a robust platform to study the molecular mechanisms underlying motor dysfunction.

PASH-1 expression reestablishes locomotor function

To determine if PASH-1 could rescue the motor defects seen in worms with the expanded 99 CGG repeats, the crawling ability of PASH-1-expressing worms was evaluated. Since the PASH-1::HA::mCherry construct was stably expressed in a single-copy 99CGG repeat-expressing worm, the red fluorescent label PASH-1 was videotaped. As in previous observations, larval worms with 99 CGG repeats showed no movement impairment, regardless of PASH-1 expression (Figure 2E, fourth bar). In adult worms, PASH-1 expression significantly improved mobility in those carrying single-copy 99 CGG repeats, with travel distances reaching approximately 5 mm, equivalent to 0.74-fold of wild-type levels (Figure 2F, fourth bar; image shown in Figure 2J). This suggests that the expression of PASH-1 mitigates motor deficits in adult worms caused by expanded CGG repeats.

PASH-1 rescues abnormal axonal morphology induced by expanded CGG repeats in motor neurons

Neuronal cellular phenotypes are implicated in FXTAS. In humans, premutation CGG alleles are associated with elevated FMR1 mRNA levels and FXTAS symptoms.^11,12,32,33 Similarly, mouse models show that CGG repeats cause neuronal dysfunction and late-onset neuropathology.^{31,34,35,36,37,38,39} Notably, reducing FMR1 mRNA ameliorates cortical neuronal phenotypes in FXTAS mice.⁴⁰

To determine whether motor defects in worms expressing 99 CGG repeats are linked to motor neuron dysfunction, we examined GABAergic neurons, which include 19 DD and VD neurons controlling forward and backward movement (Figure 3B).⁴¹ We visualized their morphology by co-expressing a mCherry reporter under a specific GABAergic unc-25 promoter in worms carrying CGG repeats of varying lengths. Commissural branches extending from ventral to dorsal sides were examined, and neurons were classified as “branched” if axonal abnormalities were present (Figures 3A and 3D).

Figure 3.

Abnormal branching phenotypes in motor neuron axons of worms expressing expanded CGG repeats

(A) Quantification of branching in VD/DD GABAergic motor neurons of adult worms expressing 0CGG, 16CGG, 99CGG, or 99CGG with PASH-1. VD/DD motor neurons were labeled using the punc-25::mCherry construct. Worms exhibiting one or more abnormal branches were classified as abnormal. The percentage of affected worms was calculated by dividing the number of worms with branching defects by the total number of worms analyzed (n values are indicated). Branching morphologies observed in 99CGG-expressing worms were classified into eight distinct categories (see Figure 3D). If a worm displayed multiple branching types, it was assigned to the most prominent morphological class. Statistical analysis was performed using one-way ANOVA followed by Tukey post hoc test.

(B) Schematic diagram illustrating approximately 18 DD/VD axons (marked in red) extending from the ventral to dorsal side in a wild-type C. elegans. Although standard anatomical references report 19 DD/VD axons, our observations using a fluorescent stereomicroscope (Leica M165 FC) consistently detected 17–18 axons in wild-type animals.

(C) Wild-type worms display normal DD/VD axons in a differential interference contrast image (left) and a corresponding fluorescent image (right). Scale bar indicates 100 μm.

(D) Representative images of VD/DD motor neurons in worms expressing 99 CGG repeats. Abnormal commissural branches are indicated by white squares. Expression of expanded CGG repeats led to a significant increase in abnormal neuronal branching. Scale bar indicates 10 μm.

Worms expressing 99 CGG repeats showed a significant increase in abnormal branching compared with wild type or 0 or16 CGG controls. Specifically, 84.7% of 99 CGG worms exhibited at least one branched DD or VD neuron (Figure 3A, fourth row), while branching was rare in control groups (Figure 3A, first, second, and third rows; Figure 3C). These findings indicate that expanded CGG repeats result in aberrant axonal morphology.

To test whether PASH-1 plays a similar role, we examined its effect on CGG-induced axonal defects. PASH-1 expression reduced branching to 38.1% in 99 CGG-expressing worms (Figure 3A, last row), significantly decreasing the frequency of neuronal abnormalities. These results suggest PASH-1 as a key regulator in counteracting CGG repeat-induced axonal defects and preserving neuronal morphology.

PASH-1 restores chemotaxis impaired by expanded CGG repeats

C. elegans has 60 ciliated sensory neurons specialized in detecting environmental cues. Among them, amphid wing “C” (AWC) neurons mediate chemotaxis toward volatile odorants such as butanone.^42,43 To assess whether expanded CGG repeats impair sensory behavior, we conducted olfactory chemotaxis assays using worms with single-copy CGG repeat insertions driven by the pan-neuronal unc-119 promoter, which includes AWC neurons.

Roughly 100 adult worms were washed and transferred to assay plates containing ethanol-diluted butanone and ethanol as point sources (Figure 4A). After 2 h, chemotaxis behavior was quantified using a chemotaxis index (CI), defined as (number at odor source – number at control)/(total number excluding origin). A CI of 1 indicates complete attraction; 0 indicates no preference.

Figure 4.

PASH-1 expression rescues olfactory defects in the C. elegans FXTAS model

(A) Schematic diagram of the olfactory chemotaxis assay. Adult animals were washed to remove bacteria and placed at the origin of the assay plate, with butanone (attractive odorant) and ethanol positioned at opposite ends. The chemotaxis index (CI) was calculated as the difference between the number of animals near butanone and those near ethanol, divided by the total number of animals that moved away from the origin.

(B) Olfactory behavior of worms expressing 99 CGG repeats showed impaired olfactory cognition. Animals expressing 99 CGG repeats in the FMR1 5′UTR exhibited impaired chemotaxis toward butanone, while the 0CGG and 16CGG lines performed similarly to wild-type worms. PASH-1 expression in the 99CGG line restored normal chemotaxis behavior, suggesting that PASH-1 is sufficient to rescue CGG-induced olfactory defects. Each data point represents the CI from one independent biological replicate. Data were obtained from independent chemotaxis assays performed as follows: wild type n = 7, 0CGG n = 5, 16CGG n = 5, 99CGG n = 5, and 99CGG + PASH-1 n = 7. All assays were conducted on separate days, with more than 100 animals tested per assay. Bars represent mean CIs; error bars represent SEM; p-values are based on one-way ANOVA.

Worms expressing 0 CGG repeats showed normal chemotaxis (CI = 0.83), indicating that high GC content alone does not impair behavior (Figure 4B, second bar). Similarly, worms with 16 CGG repeats had a comparable CI (0.79; Figure 4B, third bar). In contrast, worms expressing 99 CGG repeats showed a markedly reduced CI (0.06; Figure 4B, fourth bar), indicating impaired odor recognition when CGG repeats are expressed pan-neuronally.

Importantly, co-expression of PASH-1 with 99 CGG repeats restored chemotaxis (CI = 0.64; Figure 4B, fifth bar), suggesting that PASH-1 mitigates CGG-induced olfactory deficits. These findings reveal a protective role for PASH-1 in maintaining neuronal signaling under CGG-induced stress.

Previously, using an AWC-specific promoter, we showed that neither 0 nor 99 CGG repeats disrupted butanone chemotaxis, though 99 CGG worms failed to adapt to prolonged butanone exposure without food.⁴⁴ This suggests the chemotaxis impairment from pan-neuronal 99 CGG expression involves other neurons beyond AWC. Identifying these contributing neuronal subtypes remains an important direction for future research.

PASH-1 rebalances miRNA homeostasis disrupted by CGG repeat RNA toxicity

Expanded CGG repeats in the 5′ UTR of the human FMR1 gene contribute to FXTAS pathology through RNA toxicity. One proposed mechanism is the sequestration of RBPs, such as DGCR8, a key component of the miRNA processing complex. Our previous experiments showed that PASH-1 mitigates CGG-induced elevation of FMR1 mRNA levels. To investigate whether PASH-1 also restores downstream miRNA balance, we examined miRNA expression profiles in CGG-expanded worms and evaluated the impact of PASH-1 on miRNA homeostasis.

We first performed miRNA sequencing (miRNA-seq) to compare global miRNA expression in C. elegans strains expressing single-copy insertions of 0 or 99 CGG repeats driven by a pan-neuronal promoter. Several miRNAs showed significant differential expression (Figure 5A), among which miR-51 was upregulated in 99 CGG worms. miR-51 is conserved across species and promotes GABAergic synapse formation and GABA receptor expression,⁴⁵ processes that may be involved in FXTAS pathology. To assess functional consequences of disrupted GABAergic signaling, we performed a nose-touching assay. When the nose of the worm was gently tapped, wild-type animals expressing 0 CGG repeat exhibited immediate backward movement, whereas 99CGG worms failed to respond (Figure S4; Videos S1 and S2), indicating impaired GABAergic circuit function.

Figure 5.

Expanded CGG repeats upregulate miR-51 and downregulate glo-4 mRNA levels

(A) miRNA sequencing results comparing 0CGG and 99CGG worms under a pan-neuronal promoter.

(B) Expanded CGG repeats increase miR-51 expression. Total RNA was extracted from the indicated genotypes and analyzed using TaqMan real-time PCR. Each data point represents one independent biological replicate, with miR-51 expression normalized to control sn2343 RNA. Fold changes were calculated relative to wild-type worms. The red line indicates no change compared with wild-type worms. Worms expressing 99CGG repeats showed a ∼2.5-fold increase in miR-51 levels compared with controls (0CGG and 16CGG). Co-expression of PASH-1 in 99CGG worms reduced miR-51 expression to approximately half the level observed in wild-type controls.

(C) Expanded CGG repeats reduce glo-4 mRNA expression. Each data point represents one independent biological replicate, with glo-4 mRNA expression normalized to that of the housekeeping gene act-3. Fold changes were calculated relative to wild-type worms. Real-time PCR analysis revealed a ∼50% reduction in glo-4 mRNA levels in 99CGG worms compared with controls. Co-expression of PASH-1 restored glo-4 mRNA expression to wild-type levels.

(D) Expression levels of adt-2 mRNA, a predicted miR-51 target, remained unchanged across all genotypes. Each data point represents one independent biological replicate, with adt-2 mRNA expression normalized to act-3 and calculated as fold change relative to wild-type worms. All RNA samples used for the analyses of miR-51, glo-4, and adt-2 were extracted in parallel from five independent experiments. Bars shown in (B–D) represent mean fold changes. p-values were calculated using one-way ANOVA to compare fold changes between the indicated genotypes.

(E) A proposed model of PASH-1-mediated rescue of miRNA homeostasis disrupted by expanded CGG repeats in C. elegans. Left: In wild-type worms, PASH-1 regulates miRNA biogenesis, maintaining normal miR-51 levels and proper glo-4 expression required for GABAergic synapse function. Middle: In 99CGG-expressing worms, a 5- to 8-fold increase in FMR1 mRNA sequesters PASH-1, leading to a 2.5-fold increase in miR-51 and a ∼50% reduction in glo-4 mRNA, resulting in defective GABAergic synaptic function. Right: Overexpression of PASH-1 rescues these defects by restoring miRNA biogenesis, reducing miR-51 expression by ∼50%, normalizing glo-4 mRNA levels and alleviating synaptic dysfunction.

To validate these findings, we conducted RT-qPCR in worms expressing 0, 16, or 99 CGG repeats. mir-51 was elevated ∼2.5-fold in 99 CGG worms (Figure 5B, third bar), while 0 and 16 CGG strains were similar (Figure 5B, first and second bars). Importantly, co-expression of PASH-1 with 99 CGG significantly reduced mir-51 levels, restoring them to ∼50% of wild-type levels (Figure 5B, fourth bar). These results confirm that CGG expansion disrupts miRNA processing and that PASH-1 can reverse this effect.

We next investigated predicted mir-51 target genes using miRBase (https://mirbase.org/search/) and TargetScanWorm (https://www.targetscan.org/worm_52/), along with the published literature. Two candidate targets were prioritized: glo-4, a Rab guanine nucleotide exchange factor (GEF) required for GABAergic synapse development,⁴⁶ and adt-2, a glial-secreted metalloprotease.⁴⁷ RT-qPCR showed glo-4 mRNA was reduced by ∼50% in 99 CGG worms (Figure 5C, third bar), consistent with suppression by elevated mir-51. In contrast, adt-2 mRNA levels were unchanged (Figure 5D), suggesting glo-4 is a more responsive target.

Critically, glo-4 mRNA levels were restored in 99 CGG worms co-expressing PASH-1 (Figure 5C, fourth bar), reinforcing the role of PASH-1 in correcting miRNA-mediated gene dysregulation. To further validate this, we performed transcriptomic RNA sequencing (RNA-seq) comparing 16CGG, 99CGG, and 99CGG+PASH-1 lines. glo-4 mRNA read counts were significantly reduced in 99CGG worms compared with those in both the 16CGG and 99CGG+PASH-1 lines (Figure S5). However, we acknowledge the limitations of drawing definitive conclusions from read counts derived from single RNA-seq libraries. Accordingly, we have emphasized that our primary conclusions are based on RT-qPCR data, while the RNA-seq results are presented solely as supplementary validation. In contrast, adt-2 mRNA showed no significant changes, consistent with the RT-qPCR results. Thus, PASH-1 not only counteracts mir-51 overexpression but also rescues downstream GABAergic gene expression (Figure 5E). Together, these findings demonstrate that expanded CGG repeats dysregulate miRNA expression, specifically upregulating mir-51 and downregulating its GABAergic target glo-4. PASH-1 reverses both abnormalities, restoring miRNA pathway integrity under CGG-induced RNA toxicity. By mimicking DGCR8 function, PASH-1 restores miRNA biogenesis and gene regulation, underscoring its therapeutic potential in FXTAS and related disorders.

PASH-1 reduces FMRpolyG production and improves motility

In FXTAS and other repeat expansion disorders, RNA toxicity from elevated levels of repeat-containing transcripts is a major pathological mechanism. This toxicity is often accompanied by RAN translation, which produces abnormal proteins like FMRpolyG that aggregate in neurons. In our model, the 99 CGG repeat construct is in-frame with downstream GFP, generating an FMRpolyG-GFP fusion protein under a pan-neuronal promoter. Western blotting using an FMRpolyG antibody showed no signal in worms with 16 CGG repeats (Figure 6B, Lane 3). In contrast, a distinct 35 kDa band was detected in 99CGG worms (Figure 6B, Lane 4), consistent with FMRpolyG-GFP expression. Remarkably, co-expression of PASH-1 with 99CGG reduced the FMRpolyG signal (Figure 6B, Lane 5), suggesting that PASH-1 suppresses RAN translation or promotes protein clearance.

Figure 6.

FMRpolyG protein contributes to motor impairment in the expanded CGG repeat C. elegans model

(A) Schematic diagram of the nonFMRpolyG-expressing strain. Three stop codons (indicated in blue) were inserted immediately downstream of the non-canonical ACG start codon (highlighted in orange), effectively abolishing FMRpolyG translation while preserving the CGG repeat RNA sequence. The epitope recognized by the FMRpolyG antibody is shown as a red line.

(B) Western blot analysis of FMRpolyG expression across various CGG-expressing strains. (Top) Representative western blots showing FMRpolyG levels in the indicated genotypes. (Bottom) Quantification of FMRpolyG protein levels from three independent experiments. Each data point represents the normalized FMRpolyG signal intensity from one independent experiment, with values normalized to the β-tubulin loading control. Data are presented as mean ± SEM. Statistical significance was assessed using one-way ANOVA followed by Tukey post hoc test. ∗∗ indicates p < 0.01 compared with the wild-type strain.

(C) Crawling ability assay, as performed in Figure 2A. The 99CGG strain exhibited severe locomotor deficits, requiring approximately 292 s to exit the square, compared with around 34 s for wild-type animals. Removal of FMRpolyG expression improved crawling ability by nearly 50%, with non-FMRpolyG worms escaping in an average of 148 s. Notably, co-expression of PASH-1 in the non-FMRpolyG background further enhanced motility, reducing escape time to approximately 45 s, nearly restoring wild-type behavior. Each data point represents the time required for a single worm to exit the square. Data were obtained from 12 independent biological experiments for each genotype. Bars represent mean time; error bars represent SEM; p-values are based on one-way ANOVA.

To directly assess the contribution of FMRpolyG to neuronal dysfunction, we generated a modified 99CGG construct (non-FMRpolyG) by inserting stop codons in all reading frames between the RAN translation start codon (ACG) and the CGG repeats (Figure 6A). This construct, inserted as a single copy into the C. elegans genome, abolished FMRpolyG production, confirmed by western blot (Figure 6B, Lane 6).

Next, we tested whether elimination of FMRpolyG alone could improve motor function by performing the same experiment as Figure 2A. As expected, wild-type adults escaped a 1.5 cm square within ∼33.4 s (Figure 6C, first bar), while 99CGG worms producing FMRpolyG required ∼300 s (Figure 6C, second bar). Notably, non-FMRpolyG worms escaped in ∼150 s (Figure 6C, third bar), indicating partial recovery of motility.

To assess whether the absence of FMRpolyG affected FMR1 transcript levels, we performed RT-qPCR analyses comparing FMR1::GFP mRNA levels in worms expressing 16 CGG repeats (control), 99 CGG repeats, and the non-FMRpolyG variant (Figure S6, third bar). Transcripts in the non-FMRpolyG line showed an average 3.7-fold increase relative to the 16CGG control (set to 1), while the 99CGG line exhibited a 6.15-fold increase (Figure S6, second bar). Importantly, expression levels exceeding the commonly accepted 2-fold threshold are sufficient to induce cellular toxicity. In our crawling assay, worms harboring an expression construct that does not permit the expression of FMRpolyG recovered approximately 50% of their motility relative to 99CGG animals (Figure 6C). Videos S3 and S4 further illustrate these differences: while 99CGG worms showed severely impaired movement and lacked coordinated slow crawling, animals incapable of producing FMRpolyG still displayed an abnormal circular crawling pattern, albeit less severe than for worms that remain capable of producing the FMRpolyG-GFP protein. In contrast, wild-type worms demonstrated normal sinusoidal movement (Video S5). These findings support the interpretation that movement dysfunction is clearly due to RNA toxicity per se. Moreover, the reduced severity of dysfunction in non-FMRpolyG worms likely reflects their lower mRNA expression levels, indicating the presence of significant RNA toxicity in the 99CGG model.

Importantly, PASH-1 expression in the non-FMRpolyG strain restored motility to near wild-type levels, with an average escape time of 45 s (Figure 6C, fourth bar). This demonstrates that PASH-1 not only reduces FMRpolyG accumulation but also improves motor function. Together, these findings highlight PASH-1 as a key modulator of both RNA toxicity and RAN translation-mediated protein toxicity in FXTAS.

Glial expression of 99 CGG repeats impairs odor recognition but not movement

In FXTAS patients, astrocyte abnormalities, including altered reactivity and density, are observed in brain regions such as the striatum and cerebellum.⁴⁸ Astrocytes, a major glial cell type, are essential for brain homeostasis. To examine the effect of CGG repeat expression in glia, we used the pan-glial pmir-228 promoter to drive single-copy expression of the human FMR1 5′ UTR with either 16 or 99 CGG repeats in C. elegans (Figure 7A). Worms expressing 99 CGG repeats in glia exhibited normal crawling behavior (Figures 7B vs. 7C) but showed markedly impaired chemotaxis to the attractive odorant butanone compared with those with 16 CGG repeats (Figure 7D, second vs. third bars). Moreover, expression of 99 CGG repeats driven by a pan-glial promoter also resulted in detectable FMRpolyG production, as evidenced by the FMRpolyG-specific western blot signal in lane 7 of Figure 6B. These findings suggest that CGG RNA toxicity in glia or in combination with FMRpolyG toxicity selectively disrupts sensory behavior without impairing motor function, pointing to non-cell-autonomous effects on olfactory circuits.

Figure 7.

Glial CGG repeat expression preserves motor ability but impairs olfactory function via disruption of AWC cilia structure

(A) Schematic representation of the 99CGG construct expressed under the pan-glial promoter pmir-228, which directs CGG repeat expression specifically to glial cells.

(B and C) Crawling traces recorded over 120 s.

(B) Wild-type worms and (C) pmir-228::99 CGG worms exhibit comparable locomotion, with movement paths tracked in red lines. The dark brown region on the left indicates the E. coli lawn. Scale bar indicates 1 mm.

(D) Quantification of olfactory response, as performed in Figure 4A, reveals a significant loss of odor recognition in pmir-228::99 CGG worms compared with wild-type and pmir-228::16 CGG controls. Each data point represents the chemotaxis index obtained from one independent biological replicate. Data were obtained from five independent chemotaxis assays, each conducted on a separate day using >100 animals per genotype. Bar graphs represent mean CI; error bars indicate SEM. p-values were calculated using one-way ANOVA.

(E) EGL-4 localization in AWC neurons is not altered by glial CGG expression. The left panel shows an AWC neuron pair in the head region with the observed cell body marked by a blue square. In wild-type worms, GFP-tagged EGL-4 remains cytoplasmic during odor-seeking behavior (left), and translocates to the nucleus following prolonged odor exposure (middle), mimicking odor adaptation (odor-ignoring). In pmir-228::99 CGG worms (right), CGG repeat expression in glia results in EGL-4 remaining cytoplasmic. Scale bar indicates 10 μm.

(F) Glial 99 CGG expression disrupts AWC cilia structure. (Top) The left diagram illustrates the paired AWC sensory neurons in the head region of the worm, with their cilia marked by a blue square. According to the WormAtlas (https://www.wormatlas.org/hermaphrodite/neuronalsupport/mainframe.htm), wild-type AWC neurons exhibit distinct wing-like cilia structures. A representative image of this normal morphology is shown in the middle panel, alongside a schematic model (right). (Middle) In wild-type worms (left panel), AWC cilia retain their characteristic wing-like shape. In contrast, pmir-228::99CGG worms exhibit two distinct abnormal cilia morphologies: a hook-like structure (a) and a ball-like structure (b), indicating that CGG repeat expression in glial cells induces structural defects in AWC cilia. Scale bar indicates 10 μm. (Bottom) Quantification of AWC cilia abnormalities across genotypes. A total of 50 worms per group were scored. While wild-type worms showed no ciliary defects, glial 99CGG worms displayed abnormalities in 72% of animals, subdivided into type a (38%) and type b (34%). Statistical significance was assessed using one-way ANOVA followed by Tukey post hoc test.

To determine whether this behavioral deficit was due to disrupted neuronal signaling rather than locomotion, we examined EGL-4, a cGMP-dependent kinase whose subcellular localization in AWC neurons modulates odor seeking.^49,50,51 Cytoplasmic EGL-4 promotes chemotaxis (Figure 7E, left), whereas nuclear EGL-4 suppresses it (Figure 7E, middle). We crossed glial 99 CGG worms with a strain expressing GFP-tagged EGL-4 in AWC neurons. Imaging showed that EGL-4 remained cytoplasmic (Figure 7E, right), indicating that the odor-processing pathway was functionally intact.

We then analyzed AWC neuron morphology. Surprisingly, glial 99 CGG expression caused structural defects in AWC cilia (Figure 7F, middle panels; wild-type vs. panels a and b), which are essential for detecting volatile odorants associated with worm food. These defects likely explain the observed chemotaxis impairment. This aligns with prior findings that glia modulate neuronal structure and sensory function.⁵² By contrast, our earlier work showed that expressing 99 CGG repeats directly in AWC neurons did not affect morphology or olfactory behavior.⁴⁴ Together, these results indicate that CGG-induced RNA toxicity in glia causes non-cell-autonomous disruption of AWC neuron integrity, impairing olfactory recognition while preserving general motor control.

Discussion

Fragile X-associated tremor/ataxia syndrome is one of the leading monogenic forms of neurodegeneration in humans; however, despite its importance, suitable in vivo animal models for studying pathogenic mechanisms at the single-cell level are still lacking. We have previously shown that expression of an expanded CGG-repeat RNA in a single pair of olfactory (AWC) neurons impairs olfactory adaptation in C. elegans.⁴⁴ Olfactory defects are also noted in FXTAS, with the frequency of such defects approximately twice as high in premutation carriers as in controls⁵³; the defects were greater in carriers with cognitive impairment than in cognitively normal carriers. In the current work, we have extended our observations to motor phenotypes in C. elegans, as a model for studying the effects of the expanded CGG repeat on neuronal function. Remarkably, we find that the C. elegans ortholog of DGCR8 (PASH-1), a protein observed to correct some features of the FXTAS phenotype in cultured mouse neurons,¹³ is capable of correcting substantially all of the CGG-repeat-induced phenotypes in C. elegans (Figure 8).

Figure 8.

A model outlining the pathological effects of CGG repeat expression in neurons and the protective role of PASH-1

(A) Expression of 99 CGG repeats under the control of a pan-neuronal promoter leads to a 5- to 8-fold increase in CGG repeat mRNA and an approximately 2.45-fold elevation in FMRpolyG protein levels. This upregulation is associated with multiple downstream phenotypes, including impaired behavioral function, altered neuronal morphology, and disrupted miRNA homeostasis. These findings are consistent with toxic gain-of-function mechanisms driven by both RNA and RAN (Repeat-Associated Non-AUG) translation products in the FXTAS disease model.

(B) Co-expression of the RNA-binding protein PASH-1 significantly alleviates CGG repeat-induced dysfunctions. In the presence of PASH-1, the expression levels of CGG repeat mRNA and FMRpolyG protein were restored to near-normal levels. As a result, behavioral deficits, axonal abnormalities, and altered miRNA profiles were substantially ameliorated. These findings suggest that PASH-1 exerts a protective role by modulating the biogenesis and/or stability of CGG repeat RNA and its associated toxic translation products, offering a potential therapeutic strategy for FXTAS.

Our study demonstrates a crucial protective role for PASH-1 in mitigating the toxic consequences of the expanded CGG repeats, which are characteristic of FXTAS in humans. Central to FXTAS pathology are elevated levels of FMR1 mRNA,^10,12,54 which result in RNA toxicity, at least in part, through the sequestration of essential RBPs. Our findings indicate that PASH-1 effectively counteracts these pathological elevations of the CGG-repeat-containing RNA, consistent with previous studies showing that DGCR8 binds expanded CGG repeats with high affinity,¹³ likely preventing RBP sequestration and downstream cellular dysfunction.

PASH-1/DGCR8 is canonically essential for miRNA biogenesis through its interaction with DROSHA, converting primary miRNAs into precursor miRNAs that are further processed by DICER into mature forms. While sequestration by expanded CGG repeats would predict reduced miRNA expression, our miRNA sequencing revealed both upregulated and downregulated miRNAs in nematodes expressing expanded CGG repeats, as also observed by Sellier et al.¹³ Decreased levels align with the canonical sequestration model; however, increased levels may reflect secondary regulatory pathways or non-canonical roles of DGCR8. Indeed, the recent literature indicates DGCR8 has functions independent of miRNA processing, including DNA double-strand break repair⁵⁵ and heterochromatin maintenance through interactions with nuclear envelope proteins such as Lamin B1, KAP1, and HP1.⁵⁶ Alternative DGCR8/PASH-1 isoforms may also modulate distinct pathways through specific protein interactions.⁵⁷ Therefore, the protective effects of PASH-1 against CGG toxicity may involve both canonical miRNA-dependent and miRNA-independent mechanisms. These findings underscore the multifaceted role of DGCR8/PASH-1 in neurodegeneration and highlight its therapeutic potential in FXTAS and related disorders.

Expanded CGG repeats within the FMR1 gene are thought to underlie FXTAS pathology through mechanisms involving direct RNA toxicity^13,54 and/or indirectly, through RAN-translation to produce the toxic peptide, FMRpolyG.²³ Immunohistochemical studies revealed FMRpolyG accumulation within intranuclear inclusions that co-localize with proteins such as ubiquitin, p62, and ubiquilin 2, highlighting their pathogenic significance.^8,9,24 Despite this, the relative contributions of CGG repeat RNA toxicity and FMRpolyG protein toxicity have remained unclear. Our C. elegans FXTAS model, using a single-copy insertion of expanded CGG repeats, enabled dissection of this question. By inserting stop codons in all three reading frames upstream of the non-canonical ACG start codon, we successfully prevented FMRpolyG translation. These worms, expressing expanded CGG RNA alone, exhibited approximately a 50% improvement in motility compared with worms expressing both CGG RNA and FMRpolyG, suggesting that both mechanisms contribute to pathogenesis. The study demonstrates that PASH-1 reduces FMR1 mRNA levels and attenuates FMRpolyG production; however, the underlying mechanism remains to be fully elucidated. Potential involvement of the proteasomal or autophagic pathways is an important future direction. Similarly, a knock-in mouse model expressing expanded CGG repeats without FMRpolyG exhibited only mild behavioral deficits compared with the pronounced ataxia observed in models expressing both RNA and protein.^23,35,37

Contrasting findings from other models have clouded the role of FMRpolyG in driving toxicity. For instance, transgenic mice lacking FMRpolyG expression displayed no significant pathology, implying that CGG repeat RNA alone might be insufficient for pathogenicity.²² Furthermore, cellular models expressing mutated NGG repeats generating FMRpolyG without the RNA hairpin demonstrated severe toxicity, including intranuclear aggregation, reduced viability, and disrupted nuclear architecture. In contrast, co-expression with native RNA hairpins did not exacerbate toxicity.⁵⁸ This discrepancy may reflect differences in CGG repeat length, expression levels, or tissue specificity among the various models. In this regard, Haify et al.²⁵ failed to observe any significant behavioral phenotype in an inducible mouse model despite the presence of substantial numbers of FMRpolyG-positive inclusions, in greater than 40% of striatal neurons of 12 wo mice—much higher inclusion loads than found in humans with FXTAS. Thus, the current study lends support to the participation of both direct RNA toxicity and indirect toxicity, via FMRpolyG production, giving rise to neuronal dysfunction.

Although FXTAS was originally viewed as primarily a disorder of neurons,^6,7 more recent studies have broadened the range of neural cells that manifest dysfunction in FXTAS to include astrocytes⁴⁸ and capillary vascular endothelial cells lining the blood-brain barrier.⁵⁹ In our C. elegans model, glial-specific expression of CGG repeats did not impair locomotion, an observation that suggests a more specific involvement of neuronal dysfunction in the motor phenotypes of FXTAS. However, glial-specific expression does markedly disrupt olfactory chemotaxis, highlighting a non-cell-autonomous mechanism by which glial RNA toxicity affects at least some classes of neighboring neurons. Notably, impaired chemotaxis was linked specifically to structural abnormalities in the cilia of sensory neurons rather than defects in the olfactory signaling cascade itself, as indicated by normal EGL-4 localization. These findings align with observations in mammalian systems, where astrocyte dysfunction contributes to FXTAS pathology.⁴⁸ Moreover, glial CGG toxicity may interfere with the structural maintenance of neurons, reminiscent of prior studies showing glia-mediated regulation of C. elegans neuronal cilia,⁶⁰ and of apparent astrocyte-to-neuron dysregulation and inclusion formation.⁶¹

In addition to neural dysfunction in C. elegans expressing expanded CGG repeats under a pan-neuronal promoter, we also observed abnormal gonad development, including premature oocyte formation (Figure S7) and lifespan (Figure S8). This finding suggests that neuronal CGG repeat toxicity exerts non-cell-autonomous effects on distal tissues, such as the gonads. This is consistent with the systemic nature of fragile X-associated disorders, including fragile X-associated primary ovarian insufficiency, a condition linked to premutation CGG expansions in humans.^{62,63,64,65,66} These findings are consistent with prior reports of extra-CNS involvement in fragile X-associated disorders.⁶⁷ Taken together, these results demonstrate that CGG-induced toxicity, whether originating in glial cells and affecting neurons or arising in neurons and impacting reproductive tissues, disrupts systemic cellular and neuronal homeostasis. This underscores the broader physiological consequences of CGG repeat expansion that extend beyond classical neurological phenotypes.

In conclusion, our study provides compelling evidence that PASH-1 effectively mitigates CGG repeat-driven RNA and protein toxicities, restores neuronal integrity, and normalizes disrupted miRNA pathways in a robust C. elegans FXTAS model. These findings highlight the therapeutic potential of targeting the DGCR8/PASH-1 pathway for FXTAS and possibly other repeat-associated neurodegenerative diseases. Further exploration into PASH-1-mediated pathways in higher-order models will undoubtedly offer critical insights into therapeutic interventions for FXTAS and related disorders.

Limitations of the study

While our findings hold significant therapeutic promise, certain limitations must be noted. Translating results from a C. elegans model to humans involves challenges due to the differences in biological complexity between nematodes and mammals. However, since over 80% of all genes in C. elegans have homologs in humans,²⁶ and similar numbers in mouse models, C. elegans holds great potential for in vivo studies involving homologous gene pairs. Remarkably, one orthologous pair, DGCR8/PASH-1, substantially mitigates the effects of the expanded CGG repeat: restoring dendritic branching and neuronal viability in cultured mouse neurons and rescuing branching defects as well as multiple sensory and motor phenotypes in C. elegans. The molecular mechanisms elucidated here provide a strong foundation for future translational studies. Testing DGCR8/PASH-1 modulation in mammalian models or patient-derived neuronal cultures will be essential to validate its therapeutic potential and clinical applicability.

In this study, we used the nematode C. elegans, which exists primarily as a self-fertilizing hermaphrodite (genetically XX). Males (X0) occur only rarely (∼0.1%) under standard laboratory conditions and were not used in our experiments. Therefore, all results were derived from hermaphrodites, and sex-specific effects are not applicable in this model.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Bi-Tzen Juang (btjuang@nycu.edu.tw).

Materials availability

Strains and plasmids generated in this study are available upon request.

Data and code availability

• •All data reported in this article will be shared by the lead contact upon request.
• •All sequencing data used in this study were obtained from publicly available datasets in the NCBI Sequence Read Archive (SRA) under the accession number SRA: PRJNA1374643 or https://www.ncbi.nlm.nih.gov/sra/?term=PRJNA1374643. Mendeley Data: https://data.mendeley.com/drafts/rwmy2n7cvx.
• •Any additional information required to reanalyze the data in this article is available from the lead contact upon request.

Acknowledgments

We thank Yuji Kohara for providing yk cDNA clones and C. elegans strains provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). We thank John Wang (Academia Sinica, Taiwan) for kindly providing the original pBRC_double_attB_donor plasmid and BRC0566 strain. We also acknowledge the assistance of the Taiwan C. elegans Core Facility (CECF) for technical support. We thank Churchill Chen and Jung-Hsiang Chen (Association of NCTU Alumni) and Yan-Hwa Wu Lee (National Yang Ming Chiao Tung University) for supporting the development of the B.-T.J. laboratory. N.D.L. acknowledges support from NINDS (R01NS087544). B.-T.J. acknowledges support from the National Science and Technology Council, Taiwan (NSTC 113-2311-B-A49-003), the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan, and Yen Tjing Ling Medical Foundation (CI-114-20).

Author contributions

T.-Y.H., H.-Y.Y., N.D.L., P.J.H., and B.-T.J. conceived the study, performed the experiments, and analyzed the results. K.-Y.C., A.Y.C., and J.S. conducted the software tracking analysis. L.-S.H. and G.-Y.C. performed western blots and data analysis. T.-Y.H., H.-Y.Y., A.Y.C., J.S., N.D.L., P.J.H., and B.-T.J. contributed to the writing of this paper. B.-T.J. and P.J.H. supervised the whole project.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-FMR1polyG Antibody, clone 9FM-1B7	Sigma-Aldrich	Cat# MABN1788
beta Tubulin Loading Control Monoclonal Antibody (BT7R)	Invitrogen	Cat# MA5-16308; RRID: AB_2537819
Anti-mouse IgG, HRP-linked Antibody	Cell Signaling	Cat# 7076S, RRID: AB_330924
Bacterial and virus strains
Escherichia coli: HB101	Caenorhabditis Genetics Center	WBStrain00041075
Chemicals, peptides, and recombinant proteins
TRIzol™ Reagent	Thermo Fisher Scientific	15596018
TURBO DNA-free™ Kit	Invitrogen	AM1907
MultiScribe™ Reverse Transcriptase	Invitrogen	4311235
TaqMan™ Fast Advanced Master Mix for qPCR	Applied Biosystems	4444556
iScript™ cDNA Synthesis Kit	BIO-RAD	1708890
iTaq Universal SYBR Green Supermix	BIO-RAD	1725121
Q-PAGE™ Bis-Tris Precast Gel	Smobio	QP2310
pENTR™/D-TOPO™ Cloning Kit	Thermo Fisher Scientific	K240020SP
Gateway™ LR Clonase™ II Enzyme mix	Invitrogen	11791020
Deposited data
miRNA sequencing and RNA sequencing datasets	NCBI	SRA: PRJNA1374643
Experimental models: Organisms/strains
Wild-type Caenorhabditis elegans	Caenorhabditis Genetics Center	CGC1
pyIs500	Caenorhabditis Genetics Center	JZ500
Oligonucleotides
miR-51 TaqMan probe: UACCCGUAGCUCCUAUCCAUGUU	Thermo Fisher Scientific	N/A
sn2343	Applied Biosystems	4427975 001760
nonFMRpolyG PCR forward primer: CGGCGGCGGTGACG TAGTTGATTAAGAGGCGCCGCTGCC	IDT	N/A
nonFMRpolyG PCR reverse primer GGCAGCGGCGCCTCTT AATCAACTACGTCACCGCCGCCG	IDT	N/A
mir-228 PCR forward primer: CTGCAGCGTAAGGATATCCCGTGTGC	IDT	N/A
mir-228 PCR reverse primer:GGATCCGAGGAAAATGTCTCGCCAAA	IDT	N/A
FMR1 CGG::GFP PCR forward primer: CAGGGCTGAAGAGAACGGTA	IDT	N/A
FMR1 CGG::GFP PCR reverse primer: TTTCCGTATGTTGCATCACC	IDT	N/A
pash-1 PCR forward primer: GCTCGTCCAGTTTCAGGAAG	IDT	N/A
pash-1 PCR reverse primer: CTGTCCATCCTTCTGGCAGT	IDT	N/A
glo-4 PCR forward primer: GAGCTCTCTGCCGATGATTC	IDT	N/A
glo-4 PCR reverse primer: TGGCTTTGTGTCTCGACTTG	IDT	N/A
adt-2 PCR forward primer: ACCAATCGTCGTCCGTCTAC	IDT	N/A
adt-2 PCR reverse primer: TCGGTTTCGTCTTCATTTCC	IDT	N/A
act-3 PCR forward primer: CCCACTCAATCCAAAGGCTA	IDT	N/A
act-3 PCR reverse primer: ATCTCCAGAGTCGAGGACGA	IDT	N/A
Recombinant DNA
Plasmid: punc-119:FMR(CGG)0::GFP	This study	N/A
Plasmid: punc-119:FMR(CGG)16::GFP	This study	N/A
Plasmid: punc-119:FMR(CGG)99::GFP	This study	N/A
Plasmid: punc-25:FMR(CGG)0::GFP	This study	N/A
Plasmid: punc-25:FMR(CGG)99::GFP	This study	N/A
Plasmid: punc-119:PASH-1:HA:mCherry	This study	N/A
Plasmid: punc-119:nonFMRpolyG::GFP	This study	N/A
Plasmid: pmir-228:FMR(CGG)16::GFP	This study	N/A
Plasmid: pmir-228:FMR(CGG)99::GFP	This study	N/A
Plasmid: pBRC_double_attB_donor_BTJ	This study	N/A
Software and algorithms
WormTrackAI	This study	N/A
GraphPad Prism 9.0	GraphPad Prism Software, Inc	https://www.graphpad.com/
MetaMorph	Molecular Devices	https://www.moleculardevices.com/

Experimental model and study participant details

The C. elegans Bristol N2 strain⁶⁸ was used as the wild-type strain in this study. The JZ500 strain, expressing pAWC::GFP::EGL-4, was obtained from the Caenorhabditis Genetics Center (CGC). Worms were maintained at 20°C on nematode growth medium (NGM) plates (2.5 g peptone, 17 g agar, 3 g NaCl per liter), seeded with Escherichia coli HB101.

Method details

Plasmid construction

Punc-119::CGG repeats:GFP constructs

The 3.5 kb sequence upstream of the unc-119 start site of translation was amplified and inserted into the pAWC::FMR(CGG)₀::GFP construct,⁴⁴ replacing the pAWC promoter. The punc-119::FMR(CGG)₀::GFP fragment was amplified, with a four-base CACC extension added to the 5′ end of the forward primer, required for cloning into a pENTR/D-TOPO vector (Thermo Fisher). The pENTR:punc-119::FMR(CGG)₀::GFP construct was used in an LR recombination reaction with LR Clonase II (Invitrogen) to insert the punc-119::FMR(CGG)₀::GFP fragment into a modified pBRC_double_attB_donor plasmid,²⁷ designated pBRC_double_attB_donor_BTJ. The resulting construct was microinjected into the BRC0566 strain, which contains a phiC31 gene encoding phiC31 Integrase for inserting a single-copy gene of pAWC::FMR(CGG)₀::GFP into a precise site on Chromosome II. Successful integration restored the movement in progeny worms by replacing the defective unc-119 gene in the BRC0566 strain with a functional unc-119 gene from the pBRC_double_attB_donor_BTJ plasmid.

For constructs containing 16 or 99 CGG repeats of the FMR1 5′UTR, the sequence was obtained from pAWC::FMR(CGG)₁₆::GFP or pAWC::FMR(CGG)₉₉::GFP constructs, respectively,⁴⁴ by digestion with BamHI and EcoRI, and ligated with the pENTR:punc-119::FMR(CGG)₀::GFP construct cut with the same enzymes. The integration into the C. elegans chromosome followed the same procedure as described above.

Punc-25::CGG repeats:GFP constructs

The 1.8 kb sequence upstream of the unc-25 start site of translation was amplified and inserted into the pAWC::FMR(CGG)₀::GFP or pAWC::FMR(CGG)₉₉::GFP constructs,⁴⁴ replacing the pAWC promoter. The integration into the C. elegans chromosome followed the same procedure as described above.

Punc-119::PASH-1:HA:mCherry

The 3.5 kb sequence upstream of the unc-119 start site of translation was amplified with PstI and XbaI sites and ligated into the pPD95.75 vector cut with the same enzymes. Full-length cDNA encoding PASH-1 was amplified from yk396h9 with XmaI and KpnI sites, and ligated into the punc-119-pPD95.75 plasmid cut with the same enzymes. To add an mCherry tag downstream of the pash-1 cDNA, a PCR fusion-based approach was used, combining the HA fragment from pAWC::MUT-7:HA::GFP⁴⁹ and mCherry from pAWC::mCherry.⁵⁰ The fusion product was ligated into the punc-119::pash-1-pPD95.75 plasmid. The resulting punc-119::PASH-1:HA:mCherry construct was microinjected into wild-type worms without a co-injection marker. Transgenes were integrated into the genome using trimethylpsoralen (TMP), and integrants were screened for 100% transmission of the mCherry fluorescent signal. The integrated strains were outcrossed with wild-type worms three times to eliminate TMP-induced mutations.

Punc-119::nonFMRpolyG::GFP constructs

The short nucleotide sequence TAGTTGATTAA was inserted immediately after the non-canonical start codon ACG using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent Technologies), with a pair of primers designed using the web-based tool PrimerX (https://www.bioinformatics.org/primerx/index.htm): 5′-CGGCGGCGGTGACGTAGTTGATTAAGAGGCGCCGCTGCC.

And 5′- GGCAGCGGCGCCTCTTAATCAACTACGTCACCGCCGCCG on the plasmid pENTR:punc-119::FMR(CGG)₉₉::GFP. The resulting plasmid was sequenced to confirm the correct insertion. Single-copy chromosomal integration into C. elegans followed the procedure described above.

Pmir-228::CGG repeats:GFP constructs

Approximately 2.2 kb of sequence upstream of mir-228 on chromosome IV was amplified from C. elegans genomic DNA.⁶⁹ Primers used for amplification of the mir-228 promoter included PstI and BamHI restriction sites at the 5′ ends, respectively: 5′-CTGCAGCGTAAGGATATCCCGTGTGC and 5′-GGATCCGAGGAAAATGTCTCGCCAAA. The amplified fragment was digested with PstI and BamHI, and ligated into the pENTR:punc-119::FMR(CGG)₁₆::GFP and pENTR:punc-119::FMR(CGG)₉₉::GFP plasmids, respectively, which were cut with the same enzymes. The resulting plasmids were sequenced to confirm correct insertion. Single-copy chromosomal integration into C. elegans followed the procedure described above.

Behavioral assays

Olfactory chemotaxis

L4-stage C. elegans were grown on NGM plates seeded with Escherichia coli HB101 as food, at 20°C for 5 days. Adults were collected by washing with S-basal buffer (5 g NaCl and 50 mL of 1 M potassium phosphate, pH 6.0, per liter) into a microcentrifuge tube, followed by three washes. Worms were then placed on a 9 cm chemotaxis assay plate (10 mL of 1.6% agar in 5 mM potassium phosphate (pH 6.0), 1 mM CaCl₂ and 1 mM MgSO₄). 1 μl of 1 M Sodium azide was applied to two odorant spots during washing. Approximately 100 worms were transferred to the assay plate, with 1 μL of diluted butanone (1:1000 in ethanol) applied to the odorant spot, and 1 μL of 100% ethanol applied to the opposite spot. Worms were allowed to move for 2 h at 20°C. Olfactory chemotaxis was quantified by subtracting the number of worms at the ethanol spot from those at the odorant and dividing by the number that left the origin.

WormTrackAI: Automated quantification of crawling behavior using computer vision

Worms of specific age groups were placed in the middle of NGM plates under a fluorescent dissecting microscope, and their crawling path was videotaped for analysis. The videos were analyzed through image processing techniques for tracking of the worms’ movements. The methodology was implemented in the Python programming language utilizing the OpenCV library.

Each video contains a worm either larvae or adult. Each image frame of the videos was first converted to grayscale. Background and foreground separation was then conducted on the image. In this work, the foreground in the image is the body of the worm. After the background is subtracted, the contour of the worm is isolated. Image moment of the contour is then derived and the centroid of the contour is calculated for the worm. The worm centroid of each image frame in the video is tracked in image coordinates, and the total crawling distance is calculated by summing the position differences of sequential centroids frame by from of the video.

Quantitative real-time PCR analysis

Total RNA was extracted using TRIzol reagent,⁷⁰ followed by phase separation with 1-bromo-3-chloropropane and precipitation with isopropanol. RNA pellets were washed with 75% ethanol, air-dried, and resuspended in RNase-free water. Genomic DNA was removed using the TURBO DNA-free Kit (Invitrogen) according to the manufacturer’s instructions.

miRNA and mRNA library preparation and bioinformatic analysis

Three micrograms of total RNA were used for miRNA and mRNA library construction. All library preparation, sequencing, and subsequent bioinformatic analyses—including expression profiling, heatmap generation, and volcano plot visualization—were performed by the Taiwan Genomic Industry Alliance Inc., a commercial genomics service provider.

Quantitative real-time PCR using TaqMan probe for miR-51

Quantitative real-time PCR was conducted as described previously.⁵⁰ A TaqMan probe and primers for miR-51 (sequence: UACCCGUAGCUCCUAUCCAUGUU) were designed by Thermo Fisher. Forty-eight nanograms of total RNA were used for cDNA synthesis using Multiscribe Reverse Transcriptase (Applied Biosystems). qPCR reactions were performed in triplicate using cDNA, fluorogenic probe, and TaqMan Fast Advanced Master Mix (Applied Biosystems). Thermocycling was carried out on a QuantStudio 3 Real-Time PCR System (Thermo Fisher) with the following conditions: 95 °C for 20 s, followed by 40 cycles of 95 °C for 1 s and 60 °C for 20 s. Ct values were obtained, and relative miR-51 expression was normalized to sn2343 (Applied Biosystems). Fold change was calculated relative to wild-type worms processed on the same day.

Quantitative real-time PCR using SYBR green

One microgram of total RNA was used for cDNA synthesis using the iScript cDNA Synthesis Kit (Bio-Rad). Dye-based qPCR was performed in 20 μL reactions containing 500 nM forward and reverse primers, 2 μL of cDNA, and 10 μL of 2× iTaq Universal SYBR Green Supermix (Bio-Rad). Thermocycling was performed with 40 cycles of 95 °C for 15 s and 60 °C for 1 min.

FMR1 CGG::GFP: CAGGGCTGAAGAGAACGGTA and TTTCCGTATGTTGCATCACC.

pash-1: GCTCGTCCAGTTTCAGGAAG and CTGTCCATCCTTCTGGCAGT.

glo-4: GAGCTCTCTGCCGATGATTC and TGGCTTTGTGTCTCGACTTG.

adt-2: ACCAATCGTCGTCCGTCTAC and TCGGTTTCGTCTTCATTTCC.

act-3 (internal control): CCCACTCAATCCAAAGGCTA and ATCTCCAGAGTCGAGGACGA.

mRNA levels were normalized to act-3, and fold change was calculated relative to wild-type worms collected on the same day.

Western blot analysis

Approximately 100 μL of worm pellet was collected, resuspended in PBS containing a protease inhibitor cocktail (Sigma), and sonicated on ice (30% amplitude, 10-s bursts, repeated six times). Lysates were clarified by centrifugation at 13,000 rpm for 30 min 5× sample buffer (300 mM Tris-HCl pH 6.8, 10% SDS, 30% glycerol, 0.02% bromophenol blue) was added to the supernatant. Samples were boiled at 95 °C for 5 min, cooled, and run on Q-PAGE Bis-Tris 12% precast gels using MOPS-SDS running buffer. Proteins were transferred to PVDF membranes and probed with anti-FMRpolyG antibody, clone 9FM-1B7 (1:2000 dilution; Sigma-Aldrich) and β-tubulin antibody (BT7R, 1:500 dilution; Invitrogen), followed by incubation with HRP-linked anti-mouse IgG secondary antibody (Cell Signaling Technology). Membranes were treated with a chemiluminescent substrate, and signals were detected using a CCD camera. Signal intensities were quantified using image analysis software.

Video S1. Wild-type C. elegans with 0 CGG repeats exhibit normal GABA-dependent nose-touch responses, moving backward immediately upon stimulation.

Video S2. Worms expressing 99 CGG repeats fail to initiate backward movement in the nose-touch assay, reflecting impaired GABAergic motor responses.

Video S3. Worms expressing 99 CGG repeats exhibit severely impaired and uncoordinated crawling behavior.

Video S4. Non-FMRpolyG worms show reduced motor impairment compared with 99CGG worms, while still exhibiting abnormal circular crawling.

Video S5. Wild-type worms show normal, coordinated sinusoidal crawling behavior during locomotion.

Quantification and statistical analysis

Statistical analyses were performed in GraphPad Prism 9. Test used were described in the figure legends with sample sizes indicated.

Published: January 27, 2026