Loss of FMRP affects ovarian development and behaviour through multiple pathways in a zebrafish model of fragile X syndrome

Abstract

Fragile X syndrome (FXS) is an inherited neurodevelopmental disorder and the leading genetic cause of autism spectrum disorders. FXS is caused by loss of function mutations in Fragile X mental retardation protein (FMRP), an RNA binding protein that is known to regulate translation of its target mRNAs, predominantly in the brain and gonads. The molecular mechanisms connecting FMRP function to neurodevelopmental phenotypes are well understood. However, neither the full extent of reproductive phenotypes, nor the underlying molecular mechanisms have been as yet determined. Here, we developed new fmr1 knockout zebrafish lines and show that they mimic key aspects of FXS neuronal phenotypes across both larval and adult stages. Results from the fmr1 knockout females also showed that altered gene expression in the brain, via the neuroendocrine pathway contribute to distinct abnormal phenotypes during ovarian development and oocyte maturation. We identified at least three mechanisms underpinning these defects, including altered neuroendocrine signaling in sexually mature females resulting in accelerated ovarian development, altered expression of germ cell and meiosis promoting genes at various stages during oocyte maturation, and finally a strong mitochondrial impairment in late stage oocytes from knockout females. Our findings have implications beyond FXS in the study of reproductive function and female infertility. Dissection of the translation control pathways during ovarian development using models like the knockout lines reported here may reveal novel approaches and targets for fertility treatments.

Graphical Abstract.

Introduction

Fragile X syndrome (FXS, OMIM 300624) is a heritable intellectual disability disorder that is caused by the absence of the Fragile X mental retardation protein (FMRP) [1]. Patients with FXS display varying degrees of intellectual deficits, macroorchidism and distinct facial dysmorphisms including an elongated face and long ears. The disease mechanism of FXS has been studied extensively, both in pre-clinical models as well as clinical setting, as it is a dominant monogenic cause of autism. FMRP is expressed in all neuronal cell types, where it binds numerous transcripts implicated in autism, including those involved in cell signaling, synaptic development and function, axonal and dendritic development, and serves as a translation modulator [2]. In FXS patients, a loss of stimulus dependent translation regulation of these key mRNAs is thought to be the molecular basis of the neuronal phenotypes. However, therapies aimed at these molecular mechanisms were not found to be efficacious in clinical trials [3]. This may be explained partly based on the time of intervention, as FMRP is known to play a role in early neuronal development. It is further supported by a proof-of-concept evidence that neurodevelopmental disorders, such as FXS, may be responsive to early pharmacological intervention [4, 5].

FMRP is highly expressed in the gonads, the gametes and ubiquitously in the embryo during early development [6, 7] where it likely plays a key role in orchestrating the temporally and spatially controlled translation of maternal mRNAs, along with other RNA binding proteins (RBPs) [8]. Therefore, it is expected that a loss of this key regulator is likely to significantly impact fertility, oocyte maturation and/or early development of the embryo. This is important to study because it is probable that at least some of the neurodevelopmental defects and behavioral phenotypes seen in adults originate from this early disruption of the translation program [9]. Also important to note that primary ovarian insufficiency is a well reported Fragile-X associated condition, however it is associated with pre-mutation at the FMRP locus, and likely caused by fmr1 RNA toxicity, and not reported to be caused by a reduction or loss of FMRP [10].

Several animal models have been generated to understand the function of FMRP, including knockouts in the fruitfly, zebrafish, mouse and rat. However, there are only a few studies that focus on its role in oogenesis and early embryonic development: several in Drosophila, which reveal the role of dfmr1 in early oocyte development [11–15], and a few in mice that suggest a role in fertility [16, 17]. However, a longitudinal study investigating FMRP function at different stages of ovarian development and possible mechanisms of the observed phenotypes, has not yet been reported in a vertebrate model. The previously reported zebrafish fmr1 knockout lines mimic the neuronal phenotypes seen in FXS patients, reinforcing the idea that zebrafish is an excellent vertebrate model to study FMRP function. Therefore, in our study, we used zebrafish fmr1 knockout lines to investigate the impact of FMRP loss on reproduction and fertility at various ages. We found a delayed reproductive dysfunction phenotype in the knockout females and identified a few different molecular mechanisms contributing to this specific phenotype.

Results

Generation of the fmr1 knockout (KO) zebrafish lines

We used the standard CRISPR workflow to generate fmr1 knockout zebrafish lines [18]. Single guide RNAs were designed to target early exons (3 and 4) of zebrafish fmr1, in order to maximise the likelihood of complete loss of function (Fig. 1A). In an in vitro catalytic activity assay, both guides were able to direct Cas9 mediated dsDNA cleavage of the target DNA amplicon (Fig. S1A). Cas9-RNP complexes were assembled with both guides and used for editing the fmr1 gene in vivo. Subsequently, standard protocols were followed to identify two edited alleles (Fig. S1B), generate heterozygotes, and homozygotes. Two independent lines were maintained, each carrying a different mutant allele of fmr1: type 6 contained a 13 bp insertion and type 11 had a 4 bp deletion (5 bp deletion and 1 bp insertion), each expected to result in a premature termination of translation in exon 4 (Figs 1B and S1C). The levels of fmr1 mRNA were found to be significantly reduced (Fig. S2A), and protein levels were undetectable (Fig. S2B) in the knockout zebrafish. Both lines are henceforth referred to as fmr1 knockout (KO) zebrafish lines. The type 11 knockout zebrafish were used for subsequent studies unless otherwise indicated.

Figure 1. Characterization of the fmr1 KO zebrafish line.

(A) Location of single guide RNAs at the fmr1 locus. (B) Sequence of the edited alleles in type 6 and type 11 KO lines. (C) Imaging of the KO larvae at various time points during development (D). Quantification of the interocular distance, from Alcian blue stained larvae, showing increased spacing in the KO larvae (n = 10). (E) Relative expression levels of genes involved in craniofacial development showing a slight decrease in the KO larvae at 7dpf (n = 3).

Heterozygote crosses involving fmr1 mutant alleles resulted in progeny, which followed Mendelian ratios at birth. Even though their development appeared to be similar to wild type, the rate of survival of knockout zebrafish to adulthood (4mpf) was less than expected (16 +/− 6% S.D, as compared to 25% in four separate F2 cohorts (n = 17–32)), indicating lower fitness of the complete knockout. The fmr1 knockout zebrafish were fertile but more erratic in breeding than the age matched wildtype zebrafish beyond 9mpf. Early development of knockout embryos appeared to be marginally slower than wildtype for the first 24–48hpf (Fig. 1C). In the first two weeks, the knockout larvae were smaller and ~10%–20% had physical malformations, such as, lack of air sac, craniofacial abnormalities and jaw defects (n = 24 F2 progeny in four separate cohorts). By 3mpf, most knockout fish looked similar to wildtype, but about 20% were smaller and devoid of operculum (Fig. 1C, (n = 8 adult/group)). Minor changes in the craniofacial architecture were observed in the knockout larvae by Alcian blue staining (Fig. S3), similar to previous reports [5, 19, 20], the most prominent being an increase in inter-ocular space (Fig. 1D).

Behavioural phenotypes in the fmr1 knockout larvae

Patients with FXS display one or more of a wide range of behavioural symptoms such as increased anxiety, hyperactivity, cognitive deficits, irritability, and sleep disruptions [21]. We had previously demonstrated evidence of increased anxiety, reduced fear cognition and increased irritability in the transient fmr1 mRNA knockdown larval model [5]. However, in the fmr1 KO larvae, we did not detect a significant increase in anxiety or irritability, or decrease in fear cognition, measured by the same assays. In order to detect subtle phenotypes, Viewpoint Zebrabox and Zebralab system was used to perform automated locomotion tracking in a much large number of knockout larvae (48–96 larvae per group) at various stages during their growth. We found a robust hyperactivity phenotype at 14dpf, which was not apparent at 7dpf or 10dpf (Fig. 2A, Fig. S4A and B). However, we did not measure a robust and significant increase in the thigmotaxis behaviour (a measure of anxiety driven by a novel environment) in the larval assays even at 14dpf. However, in the KO larvae we observed a much higher rate of locomotion response in response to a light-to-dark or dark-to-light transition (Fig. S4C and D), which is supportive of an increase in stimulus driven anxiety in these larvae.

Figure 2. Behavioural phenotypes and molecular changes in the KO larvae and adult zebrafish.

(A) Locomotion tracking in individual larvae showing hyperactivity in the KO larvae at 14dpf (n = 90/group). (B–D) Behavioral assays in adult zebrafish (6-9mpf) (B). Light dark preference test: Percentage time spent in dark zone (C). Novel tank test. Percentage time spent in bottom zone. (D) Open field test: Percentage time spent in outer zone. Caffeine treated fish were used as a positive control for increased anxiety (n = 6/group). (E) Relative expression levels of various groups of genes as described in the text (n = 4/group).

Since we found a difference in the behavioural phenotypes exhibited by the KO larvae (ours as well as previously published reports [20, 22]), and the larvae with a transient mRNA knockdown [5, 23], we tested if this was due to a compensation from the upregulation of other fmr1 like genes fxr1 and fxr2. While we observed a robust decrease in fmr1 mRNA levels at 7dpf in the KO larvae, we did not observe a significant increase in the expression of either fxr1 or fxr2 (Fig. S2A). And, as previously reported, we observed a decrease in the levels of bdnf (neuronal development), and moderate but reproducible decreases in fosl2, and sox9a (contributing to the mild craniofacial deformities) (Fig. 1E). Among all of the other markers of early neuronal development, the only other significant changes observed were a robust increase in GFAP, a marker of astrocytes and a decrease in olig2, the oligodendrocyte marker (Fig. S5A).

A major theory in the FXS field postulates that mGluR5 upregulation is the key basis of the pathomechanism of FXS [24]. In the absence of FMRP, which regulates the translation of mGluR5, the mGluR5 protein level and the downstream signalling cascade is upregulated, leading to stimulus independent signalling and excitotoxicity. Therefore, to test this hypothesis, we measured the levels of mGluR5 and other members of the signalling cascade (p-ERK and p-Akt) in the KO and matched wildtype larvae, and saw no significant increase in these markers (Fig. S5B). This is in contrast to what we observed when fmr1 mRNA was transiently knocked down [5], where mGluR5 as well as the downstream signalling was upregulated. In the case of the knockout, it is possible that there may be a compensatory pathway of mGluR5 translation regulation in zebrafish that is yet to be characterized, or localized changes in expression or activation, which were not captured in these experiments.

Behavioural phenotypes and altered gene expression in the fmr1 knockout adult zebrafish

The behaviour phenotypes and brain-specific gene expression were assessed in the adult KO zebrafish (>6–12mpf, males). Adult behaviour was recorded and analysed using AnyMaze software. Three established paradigms were used [25]. The light–dark preference test, which relies on the innate preference of adult zebrafish for the dark environment. Any increase in anxiety results in the fish spending more time in the dark side. The “novel tank diving” (NTD) test which relies on the zebrafish’s innate propensity to dive, freeze, and limit exploration in novel settings. Increased anxiety results in increased amount of time spent in the deeper zones. The open field test (OFT) to measure thigmotaxis, which is a typical method for evaluating exploratory behaviour and general activity in zebrafish. Increased anxiety or intellectual deficits will result in reduced exploration.

In each of these tests, the KO fish spent more time in the dark zone (Fig. 2B), the bottom layer (Fig. 2C) and the outer zone (Fig. 2D), much like caffeine treated wild-type fish, which demonstrates an increased level of anxiety and a reduced tendency to perform exploratory activity in the KO fish (an indirect indicator of intellectual abilities). However, very similar to what was observed in the KO larvae, we saw no upregulation in the mGluR5 protein levels, or activity as measured by p-ERK and p-Akt levels in the brain lysates made from adult KO zebrafish (Fig. S6A–C). It is possible that these changes occur in specific regions or cell types in the brain, and are therefore not measurable in an assay performed on the whole brain lysate. These results suggest that the observed phenotypes may be mediated by changes that occur independent of the mGluR5 pathway.

We carried out gene expression analysis in the brain tissue of KO and age matched wildtype fish, for three classes of genes a) FMRP targets (mRNAs directly bound by FMRP or mRNAs whose polyadenylation profile/translation is altered in the absence of FMRP) b) genes associated with autism c) genes whose expression has been reported to be changed in other FXS models or patients (Table S1). We found significant changes in neuronal markers indicative of increased astrocytosis (GFAP), reduced neuronal number and cognition (rbfox3 and camta1a), BBB integrity (aqp4) and significant decrease in the inhibitory neurotransmitter receptor GABA-A [2] (gabarg2). We also found a reduction in neuronal translation control factors (CPEB1 (zorba) and the ELAVL family protein, elrA). Consistent with recent reports suggesting that Huntingtin (htt) is a direct mRNA target of FMRP [26], we found a decrease in htt levels in the KO zebrafish brain, and a corresponding change in several genes involved in mitochondrial dynamics indicative of reduced fusion (Fig. 2E). A decrease in protein levels of Mfn2, Opa1 and Zorba was also detected in the KO brain tissue (Fig. S6D–F). This mitochondrial dysfunction and persistent astrocyte activation in the FXS brain, could together be the basis for impaired neuronal development and function manifested in the form of altered behavioural responses in these KO zebrafish.

In contrast to the drastic decrease in the expression of the GABA-A receptor gabarg2 observed in the KO male brain (old fish (Fig. 2E) and young fish (Fig. 3A)), the expression levels of this receptor were unchanged in the young female KO brains (Fig. 3B). In the sexually mature adult brain, GABA acts as an excitatory neurotransmitter for GnRH neurons [27, 28] eliciting the release of GnRH, which subsequently promotes release of gonadotropins (LH and FSH) from the pituitary. In addition, in the brains of sexually mature female KO fish, we detected a reproducible increase in the expression of the neuroendocrine hormones gnrh3, and fsh (Fig. 3C). These findings support the idea of altered hypothalamic–pituitary control of reproductive development in the female KO fish, similar to that reported in the female KO mice [16].

Figure 3. Altered neuroendocrine signalling in fmr1 KO.

(A) Relative gene expression of gabarg2 and rbfox3 in brain tissue from male (A) and female (B) zebrafish. (C) Relative expression of gnrh3, fshb and lhb mRNA in 5-6mpf KO female brain tissue relative to wildtype (n = 3/group).

Loss of FMRP appears to promote increased oogenesis in juvenile zebrafish

The cytoplasmic polyA element binding proteins (CPEBs), regulate cytoplasmic polyadenylation (CPA) and exert tight spatial and temporal control of translation of maternal mRNAs in the oocyte during oogenesis, oocyte maturation and in the early embryo until zygotic transcription is initiated [29]. We investigated the impact of FMRP loss on the zebrafish oocyte CPEB zorba, the embryonic CPEB and Zorba target, elrA, and their targets, using our KO lines. We carried out this analysis in fish of three different ages: juvenile—1mpf; young adult—5–6mpf and old adult—9–12mpf, to cover various stages of ovarian development. We measured the polyadenylation status (considered a direct read-out of translation) using a PAT-PCR assay (polyadenylation tail length PCR assay, Fig. S7) as well as the gene expression profile of the appropriate targets at these time-points.

In the ovaries of 1-month old juvenile fish, where ovary differentiation has just begun and oocytes are in Stage IB, a significant increase in the mRNA levels zorba and of several key genes was observed. These included genes involved in oocyte maturation, meiotic entry and progression such as dazl (an RNA binding protein that regulates translation, required for germ cell proliferation and germline stem cell (GSC) establishment); cmos (a Serine/threonine protein kinase required for oocyte meiosis and microtubule organization); buc (the germplasm organiser and homolog of Oskar from Drosophila, regulates the number of germ cells and proper localization of GC genes nanos, dazl and vasa); ccnb1 (the cyclin B1 gene, a component of maturation promotion factor (MPF), required for proper meiotic progression); and RINGO/spdy, an atypical activator of cyclin-dependent kinases and necessary for CPEB-directed polyadenylation (Fig. 4A). Overexpression of dazl and buc are likely to result in excess GC formation, in unexpected locations, mimicking the observations in the dfmr KO flies [30, 31]. We did not observe a significant difference in the polyA profile of the zorba targets at this age, indicating no major changes in translation (Fig. 4B).

Figure 4. Gene expression profile in juvenile and young adult zebrafish (1mpf and 6mpf).

(A and C) Relative expression levels of genes involved in CPA, meiotic entry and progression in ovaries (1mpf, A) and oocytes (6mpf, C) n = 3 fish/group. (B) PAT assay for zorba, elrA, cmos and ccnb1 in the same sample as in A.

Adult zebrafish have asynchronous ovaries, containing follicles (oocytes) of all stages of development, each stage defined according to size and morphology [32, 33]. In stage III/IV oocytes from young female fish, the expression of the same markers was neither up nor down regulated in the KOs relative to wildtype, indicating that those oocytes that progressed through maturation are similar to those in wildtype, thus explaining the ability of these fish to produce viable embryos (Fig. 4C).

Altered oocyte morphology and distribution in apparently normal young knockout females

In order to assess the distribution and morphology of follicles in a young adult female, we undertook histological analysis of ovarian sections in young adult (5–6mpf) females at the same stage in the breeding cycle. H&E stained ovary sections revealed a very different distribution of the different stages of oocytes in the KO, with fewer stage I and more stage III/IV oocytes (representative images of the sections are shown in Fig. 5A and B, Fig. S8A and B (in colour), and an assembled image of the ovary cross-section is shown in Fig. S9). Additionally, in the KO, a number of large follicles containing fused structures (instead of yolk granules) characteristic of atretic follicles were present, and they lack a well-define zona radiata (Figs. 5B, 8B and S10A). The mitochondria in these KO oocytes, appeared to be less in number (lower intensity) with a more diffuse architecture as compared to the distinct Balbiani body-like structure observed in wildtype (Fig. S10B). While these fish ovulate normally at this stage, their ovaries appear to contain a reduced reserve of good quality oocytes, and resemble the ovary of a much more aged fish.

Figure 5.

Histological analysis of H&E stained ovarian sections from wildtype (A) and KO (B) zebrafish (5mpf). PV: PreVittelogenic; EV: Early Vittelogenic; LV: Late Vittelogenic; AtF: Atretic follicles.

Loss of FMRP drastically alters cytoplasmic polyadenylation in late stage oocytes from old females

In oocytes from old FMRP KO females, a drastic reduction in the level of zorba mRNA was detected (Figs 6A and 7A). And a consequent and untimely increase in both the level and polyA tail length of elrA mRNA was observed in the KO oocytes (Fig. 6A). However, the level and polyA tail length of elrA post fertilisation, appear to be similar to that of wildtype (Fig. 6A and Fig. S11B). These results are contradictory to the findings from flies, and from juvenile zebrafish ovaries, where FMRP loss results in Zorba upregulation. To validate our findings, we assessed the polyA profile of the mRNA targets of Zorba and ElrA, to see if their polyadenylation was also changed consistently in the oocytes from old KO females. As expected, several targets of Zorba show an altered polyA profile in the KO oocytes, including buc, ccnb1 and chtf8 (the orthologue of Drosophila fs(1)K10, a protein required for dorsal-ventral specification and required for egg as well as embryo polarity). Similarly, an ElrA target, hnrnp1ab, a protein that in turn controls the translation of several embryonic mRNAs, was also found with increase polyA tail length (Fig. S11C). Unexpectedly, we found a significant decrease in the polyA tail length of an oocyte tubulin gene tubb4b, which is not a known target of Zorba or ElrA. Further, we confirmed the drastic decrease in protein levels of Zorba as well as Tubulin, in these oocytes from old females, by IHC as shown in Fig. 6C and Fig. S12C (in colour).

Figure 6. Altered polyadenylation profile of transcripts in oocytes from old knockout fish.

(A) PAT assay for zorba and elrA from stage III/IV eggs, 0hpf and 2 hpf embryos. (B) PAT assay in Zorba target genes (buc, ccnb1) and other transcripts controlled by CPA (tubb4b (tubulin)) in the same samples as in A. (C) IHC with the α-tubulin and orb (Zorba) antibodies in stage III/IV oocytes from older (9–12mpf) female KO and wildtype fish.

Figure 7. Effect on FMRP loss on gene expression in oocytes from old female zebrafish (9–12mpf).

(A) Genes involved in PGC specification and differentiation, GSC proliferation, and meiotic entry and progression (B). FMRP target pum2, its target RINGO/spdy and the embryonic PABP. (C) Three different tubulin genes expressed in oocytes. (D) Expression levels of odc1, a rate-limiting enzyme in the spermidine synthesis pathway in oocytes from zebrafish of different ages, (E). Relative expression of mitochondrial markers. (F) IHC with the Tom20 antibody to visualize mitochondrial morphology and intensity in the same oocytes.

FMRP loss affects the expression of key genes involved in oocyte maturation

In addition to the polyA changes, a significant decrease in the mRNA levels of several key genes involved in oocyte maturation, meiotic entry and progression such as dazl, cmos, buc and ccnb1 was also observed (Fig. 7A). The magnitude of reduction seen in the FMRP KO oocytes from the 9-12mpf fish is even more than that observed in oocytes from 18mpf wildtype fish (Fig. S13A) and hence not just a consequence of aging. Additionally, the FMRP target pum2 mRNA, an RNA binding translation regulator, was found to be increased (Fig. 7B). And the Pum2 mRNA target RINGO/spdy which is not regulated by CPA, and is important for meiosis resumption, was also found to be decreased. We measured a drastic decrease in the levels of the embryonic polyA binding protein (ePAB, pabpc1l), the predominant cytoplasmic PABP expressed in oocytes and early embryos which prevents dead-enylation of mRNAs (Fig. 7B). All of these genes are required for the release from MII and final maturation [34–36]. The levels of several tubulin genes expressed in oocytes were also reduced suggesting an impact on the cytoskeleton (Fig. 7C). Such drastic changes are likely to severely impact translation and proper localization of mRNAs in the oocyte, thus impacting fertilization and subsequent development of the embryo.

To test whether the effect of FMRP loss was the reason for these major changes, we injected an FMRP antibody into stage III/IV oocytes collected from a wildtype female fish. The antibody is expected to sequester FMRP away and mimic the KO. As seen with the KO, the level of zorba mRNA was reduced in these oocytes, but not in those injected with a mock antibody (Fig. S13B). This suggests that FMRP is likely to stabilize the zorba mRNA and prevent its degradation in Stage III/IV oocytes.

Taken together, these results suggest that FMRP loss affects the transcriptome of oocytes in very different ways in fish of various ages (with oocytes at various stages), with a marked increase in meiotic genes in juveniles, to an unaltered profile in a young adult, all the way to a drastic downregulation of the same markers in an older adult female.

FMRP loss has a severe impact on mitochondrial health in oocytes

Polyamines have been shown to play an important role in reproductive function, among many other processes. Spermidine levels decrease with age, and directly result in poor fertility in aged female mice [37]. We measured the level of the key enzyme involved in spermidine synthesis (odc1) and found it to be reduced in the knockouts at all ages (Fig. 6D), and most drastically in the oocytes from older fish. Reduced spermidine results in impaired mitophagy, and increased apoptosis, and may account for the increase in atretic follicles in the KO fish right from 5-6mpf (Fig. 5B). In the oocytes from old KO females, we found a decrease in several genes important for mitochondrial biogenesis and function such as pgc1α, tfam, sod2, porin; and a significant decrease in the fusion genes mfn1 and mfn2 as well as an increase in the fission marker drp1 (Fig. 7E). Further, we observed much weaker and more diffuse signal for mitochondrial staining (IHC with a Tom20 antibody) in oocytes from the KO fish (Fig. 7F and Fig. S14F (in colour)) suggesting a severe impact of FMRP loss on oocyte mitochondria, which is likely to impact oocyte and subsequently embryo quality.

Discussion

In this study, we created and characterized fmr1 KO zebrafish lines, which model certain aspects of human FXS behavior such as hyperactivity and anxiety, in the juvenile and adult stages. These behavior phenotypes are accompanied by and rooted in the altered expression of genes that regulate neuronal function. In addition, we have also catalogued differences in ovarian development and oocyte maturation in the fmr1 KO females, and report on at least three molecular mechanisms that could contribute to these defects. The first is evidence of increased levels of genes that encode the neuroendocrine hormones that promote oocyte maturation, in the brain of female KO zebrafish. The second is drastic differences in the expression of key genes responsible for PGC specification, GSC proliferation and meiotic progression, and an altered distribution of oocytes at various stages in the KO zebrafish. And finally, we show proof of mitochondrial dysfunction in both the brain and the oocytes of KO zebrafish, which possibly contributes to neuronal cell death, and increased apoptosis and poor quality of oocytes in the older KO females.

FXS has been modeled in zebrafish previously using Morpholino knockdown [19] or a DNAzyme knockdown [5], a mutant line generated by ENU mutagenesis [22, 38] and by CRISPR mediated gene editing [20], with each study reporting specific phenotypes. In general, stronger phenotypes were observed with the transient knockdown methods including hyperactivity, altered anxiety, and craniofacial defects in the larvae, with a measurable impact on the mGluR5 signaling cascade. Subtle neuronal phenotypes such as altered brain-wide auditory networks [39], changes in synaptic density and locomotor activity [40] and learning and memory deficits [20] have been reported in the stable KO lines (hu2787, and the CRISPR mutant) but none report evidence of mGluR5 signaling perturbations. Only a few studies report phenotypes in older fish which include increased anxiety and precocious shoaling at 28dpf [41], hyperactivity, anxiolytic-like behavior and impaired learning accompanied by enhanced LTD in adults [38]. Consistent with the literature, we observe no robust phenotypes in 7dpf larvae, and only minor gene expression changes in olig2, a marker of oligodendrocytes and GFAP, a marker of astrocytes. This is consistent with the hypomyelination observed in infants with FXS [42].

However, in older larvae (14dpf) and in adults, we detected robust behavioral changes consistent with increased anxiety, reduced exploration and impaired cognition. FMRP deficient immature neurons in KO mice, have previously been shown to have reduced Huntingtin (htt) mRNA and protein levels, which results in impaired mitochondrial function, fragmented mitochondria and defective dendritic maturation [26]. We found a similar decrease in htt mRNA and mitochondrial fusion genes which could be the basis of improper neuronal maturation. FMRP is known to play an important role in astrocytes as well, and neurons cultured on FMRP deficient astrocytes show abnormal dendritic morphology [43]. Mitochondrial dysfunction in astrocytes is known to result in reactive gliosis (increased GFAP) and increased neuronal cell death (reduced rbfox3) [44], both of which we observe in the brain from the adult KO zebrafish. These molecular and behavior changes in the adult correlate well with each other, and with changes observed in the mouse model as well as humans with FXS, and have been demonstrated for the first time in the adult FMRP KO zebrafish.

We did not find a convincing rationale for the differences in observed phenotypes in larvae, between the transient knockdown and the stable knockout of FMRP. We or others [45] found no obvious compensatory gene expression changes. But it is well known that FMRP plays a critical role during early development, where temporal and spatial control of maternal mRNA translation is critical. Therefore, we hypothesized that in the knockout line, compensatory mechanisms at the level of translation could be activated early on to ensure survival through the MZT in the absence of FMRP, which may not continue to be at play in the adult brain or gonads. For example, even though we observed drastic changes in the expression of zorba mRNA, and polyadenylation of its targets such as elrA and hnrnp1ab in the oocytes, we did not observe this to be carried over post fertilization, with both elrA and hnrnp1ab levels and polyA profile remaining similar to wildtype (Figs 6A and S9C).

FMRP is known to be expressed in the ovaries of flies, mice, rats and humans [6, 46], but there are not many studies delving into the exact function of FMRP during oogenesis. In humans, both males and females with FXS have been reported to be fertile and capable of reproduction [47, 48]. But other than reports suggesting it is possible, there have been no detailed studies on ovarian development, ovarian reserve and the actual window of fertility. Premature ovarian insufficiency (POI) and diminished ovarian reserve (DOR) have been studied extensively in premutation carrier females (PM), of whom 25% are reported to be infertile [10]. It is believed that the increased levels of the toxic fmr1 RNA in PM carriers is more detrimental to ovarian dysfunction.

The most amount of work has been carried out in Drosophila, where FMRP is implicated in PGC proliferation, GSC specification and in the regulation of translation, especially via the microRNA pathways [11, 12]. One report in the mouse Fmr1 KO mouse showed that these mice exhibited an earlier decline in fertility, premature recruitment of follicles due to elevated mTOR activity [17]. Our study is the first to investigate reproductive and ovarian phenotypes in FMRP KO zebrafish, and our findings are very similar to that seen in the mouse model, in terms of increased oogenesis in the juvenile stage, premature maturation of follicles in the young adult resulting in the depletion of ovarian reserve, followed by a rapid and early deterioration of oocyte quality in the older females. At a molecular level, we sought to connect the observed phenotypes to the cytoplasmic polyadenylation machinery and its regulation via FMRP. The CPEB1 protein is known to be critical for oogenesis and oocyte maturation in flies, zebrafish and mice [49–53]. A functional interaction between FMRP and CPEB1 is also well reported in multiple contexts, where both proteins regulate each other’s activities, predominantly in an antagonistic fashion [54, 55]. The strongest evidence we see of the FMRP-Zorba functional interaction in our model is in late stage oocytes from older females, where both in the KO and in wildtype oocytes injected with an FMRP antibody, there is a reduction of zorba mRNA and protein levels, as well as altered polyA profiles of known Zorba targets (in the KO). Even though this is the opposite of what is seen in the Fmr1 KO mouse brain cortex, and in FMRP KO cell lines, it is possible that this is true in this specific context. These oocytes are of poor quality and resemble those from a much older wildtype female, and studies have demonstrated a decline in both CPEB1 and FMRP levels in aged females, although a causal link has not been established [6, 53].

The KO fish also show altered expression of genes involved in germline specification, meiotic entry and maturation, explaining the defects seen the distribution of oocytes in various stages as well as the quality of oocytes from older females. The novel finding is the evidence of increased neuroendocrine gene expression in sexually mature KO fish, much like in the mice, which could have an additive effect in promoting the precocious maturation of oocytes and the subsequent depletion of the ovarian reserve. However, it remains to be demonstrated in this specific KO, exactly what the downstream implications of this increased hormone gene expression are, and if CPEB1 or other RBPs are involved.

Finally, we also see a strong detrimental impact of FMRP loss on mitochondrial function in both oocytes and in the brain, as has been reported in several model systems and contexts. While in the brain, the mitochondrial impairment is due to a reduction in Huntingtin, in oocytes, we connect these findings to a downregulation of the spermidine synthesis pathway. The mechanism by which FMRP regulates this pathway remains to be uncovered.

Our study in the novel fmr1 KO zebrafish lines expands on the array of neuronal as well as reproductive FXS phenotypes studied in zebrafish, and strongly supports the idea of using this model for identification of new pathways for therapeutic interventions, especially based on the leads identified here. The zebrafish model is much better suited to large scale analysis of reproductive and behavioural phenotypes than the mouse model, because of the ease of obtaining large number of animals for experimentation, the access to early developmental window and the availability of a large number of transgenic reporter lines and genetic methods. In future studies, it may be interesting and important to determine whether mGluR5 mediated upregulation is present using detailed neurophysiological studies, and thereby reveal alternate pathways if they exist. We also catalogued the reproductive phenotypes in the knockout females, and identified certain changes in gene expression which may be causal. However, detailed studies will be required to identify the all targets of FMRP in oocytes at various stages, and then to map their functional contributions to a successful oocyte maturation program.

Materials and methods

Animal ethics procedures and approval

All experiments with zebrafish were done in a CCSEA-approved zebrafish facility at Dr Reddy’s Institute of Life Sciences (1100/po/Re/s/07/CPCSEA) in Hyderabad, India. The facility also has US-NIH OLAW assurance (F22-00539). All procedures and protocols were reviewed and approved by the Institutional Animal Ethics Committee (Protocol approval DRILS/IAEC/AS/2019-1). The “Guidelines for Experimentation on Fishes, 2021” published by CPCSEA was used as a reference.

Zebrafish culture, maintenance and breeding

The origin of the zebrafish used was a group of Indian wild-caught fish obtained from a commercial breeder. These fish were grown and bred in-house over more than 3 years. The fish were maintained in a semi-automated housing system (ZebTEC rack with active blue technology, Tecniplast, Italy). This system circulates filtered and aerated water to all fish tanks and maintains a stable pH of 7.0, a temperature of 28°C, and conductivity of 500 μS. A photoperiod of 14 h light and 10 h dark was followed. The fish were fed three times every day with either dry or live feed (pellets or live brine shrimp hatched in-house) with live feed at least once daily. Adult fish between three months and twelve months of age were bred for experiments. Breeding tanks (1.7 L) were used to set up pairs of male and female fish on the evening prior to the breeding day. Zebrafish breeding is responsive to photoperiodic cues and they lay eggs or sperm at first light. Typical breeding pairs in our facility produce an average of 300 eggs with an 80% fertilization rate. Unfertilized or non-viable eggs were discarded and fertilized eggs were transferred into a fresh Petri dish, washed at least two times in system water and maintained in a 28°C incubator.

sgRNA synthesis

Target regions for CRISPR mediated editing for fmr1 were chosen using Benchling software in exons 3 and 4. Single guide RNAs were synthesized in-house using standard protocols as described previously [56]. Briefly, sgRNA was ordered as single stranded DNA with the addition of T7 promoter at 5’end and tail oilgo sequence at 3’end (Table S2). DNA template for sgRNA synthesis was generated by standard overlap PCR with the sgRNA and tail oligos, and used for in-vitro transcription using the HiScribe™ T7 High Yield RNA Synthesis Kit (NEB). Finally, sgRNA was treated with DNase (NEB) and purified using the RNA Clean and Concentrator kit (Zymo Research) according to manufacturer’s instructions. sgRNA concentration was measured using a Biospectrophotometer (Eppendorf) and integrity was checked on the urea PAGE.

Microinjection and generation of FMR knockout lines

CRISPR/Cas9 mediated knockout generation was performed as described previously [18]. Briefly, one cell stage zebrafish embryos were injected with 5 nl of Cas9-sgRNA mix (400 ng of Cas9 protein and 0.2–0.5 μg of sgRNA in5 μl of 10 mM Tris pH 8.0). After 24hpf, dead and abnormal embryos were removed and remaining were raised to adulthood as F0 founders. Sexually mature F0 founders were outcrossed and F1 offspring were screened for indels by tailclip PCR using the standard heteroduplex mobility assay (HMA) with genotyping primers (Table S2). Positive F1 offspring were raised to adulthood, F1s of the same type were interbred to generate F2, and homozygotes were sequenced to identify the edited allele. For genotyping, 24hpf embryos or tail fin clip was heated at 95°C for 30 min in 45 μl of 50 mM NaOH solution and followed by neutralization with 5 μl of 1 M Tris pH 8.0. After a short spin, 0.3 ul was used as input for a 5 ul PCR reaction. PCR reaction was set up with 2X Q5 mix and genotyping primers. PCR product was run on 10% NATIVE PAGE.

Locomotion and behaviour analysis

Locomotion and behaviour analysis in larvae was performed as reported previously for manual recording [5]. For automated behaviour analysis, the Zebrabox system and Zebralab software were used. Single larvae (7, 10, 14dpf) were gently placed in a 24 well plate (in 1 ml of E3) 1 h before the experiment and placed inside the Zebrabox for acclimatization for 10 min. For the light dark locomotion assay, locomotion was recorded over 10 min dark, followed by alternative light/dark cycles of 3 min each. The integration period was set to 1 min; total distance was calculated for each minute. The data obtained was analysed and graphed using Microsoft Excel, and reported as reported as mean ± SEM (n = 60 (7dpf), n = 15 (10dpf), n = 90 (14dpf)).

Behaviour assays in adult male zebrafish were performed manually, as reported in [57] with minor modifications described below, and analysed using AnyMaze software. The light dark test was conducted in an aquarium of size 220 × 210 × 140 mm, separated into two chambers, one of which was covered with the white paper and the other with matte black paper covering it on all sides of the tank. A camera pointed towards the light chamber with the LED light panel was used to record for 10 min at 30 frames per second covering the whole area of the tank. The tank was filled with 5 litres of water that included 1.2 g of sea salt for conductivity. The fish were acclimatized in the behaviour room for 1 h, and then introduced into light dark box and recorded for 10 min. The percentage time spent in the dark zone was measured. The novel tank was conducted in a tank apparatus consisting of a narrow tank divided horizontally into a top and bottom zone. A camera pointed vertically towards the chamber was used to record for 10 min at 30 frames per second. The tank was filled with 5 litres of water that included 1.2 g of sea salt for conductivity. The fish were acclimatized in the behaviour room for 1 h, and then introduced into the test tank and recorded for 10 min. The percentage time spent in the bottom zone was measured. The open field test (OFT) was conducted in the OFT Box (220 × 210 × 140 mm) with a defined inner zone and outer zone. The fish were acclimatized in the behaviour room for 1 h, and then introduced into the test tank and recorded for 10 min. The percentage time spent in the outer zone was measured. The data from these experiments were analysed using Graphpad PRISM and reported as mean +/− SEM (statistical test One-way ANOVA).

Gene expression analysis: RNA extraction and quantitative real-time PCR

RNA was isolated from different tissue (7dpf larvae, adult brain, oocytes and ovary) using Trizol (NEB) and purified using the RNA Clean and Concentrator kit (Zymo Research) according to manufacturer’s instructions. For tissue (brain and ovary) homogenization was done with RNase free plastic pestle and larvae were homogenized using a 27G syringe. cDNA was made using 500 ng RNA with PrimeScript RT Reagent Kit (TaKaRa). Real time PCR was set up using TB Green Premix Ex Taq II (Tli RNase H Plus) (TakaRa) in a Quantstudio5 machine. RNAPD (polr2d) was used as the reference gene. Sequences of all the primers used in the study are listed in Table S2. The changes in RNA expression were quantified by the ΔΔCT method, normalized to the internal control, and then, in certain experiments, to the control group for each experiment. The results from three or more independent experiments were averaged and are reported, with error bars representing S.E.M. Statistical analysis (t-test) was performed using Graphpad PRISM software.

Protein measurement by western blot

Zebrafish larvae or adult brain were lysed in RIPA lysis buffer with protease inhibitor cocktail (10 μg/ml each aprotinin and leupeptin). Protein concentration was determined by BCA method. Approximately 30 μg protein was loaded on 10% SDS-PAGE and then transferred to PVDF membrane (Millipore) followed by blocking, primary (1:1000–5000) and secondary antibody (1:5000–10000) incubation. Antibodies used are listed in Table S3. To probe for multiple proteins, the membrane was cut at the appropriate size marker. Signals were detected with ECL western blotting substrate (TaKaRa) and image was acquired using Azure Biosystems. Images were quantified by using Image J. Alphatubulin was used as the reference gene for protein measurements.

Immunohistochemistry

IHC was performed as described in [58]. Briefly, stage III oocytes (~0.4–0.7 mm diameter, [32]) were collected from adult zebrafish (n = 3/group) after dissection, fixed in 4% PFA overnight and then washed for 10 min (5 times) with PBST. Oocytes were dehydrated and then rehydrated with methanol followed by blocking (10% horse serum in PBST) for 4 h at room temperature. They were incubated with primary antibody (1:200 in blocking solution) for 3 days at 4°C and then washed 5 times (20 min) with PBST at room temperature and incubated with the appropriate Alex Flour-488 conjugated secondary antibody for 2 days at 4°C. Oocytes were washed 5 time (20 min) with PBST at room temperature and then mounted in mounting media. Images of 5–10 oocytes/n were taken using EVOS cell imaging system.

PCR polyadenylation tail (PAT) assay

PCR polyadenylation tail (PAT) assay was performed as described previously [59]. Briefly, 250 ng RNA after DNase treatment was heat denatured in the presence of 56 ng of phosphorylated oligo (dT) in 7 μl volume at 65°C for 5 min and immediately placed at 42°C. 13 μl of pre warmed master mix (4 μl of 5 X RT buffer, 2 μl of 0.1 M DTT, 2 μl of 2.5 mM dNTPS, 0.1 μl of 100 mM ATP and 2.5 μl of T4 DNA ligase (40 U/μl) and 2.4 μl of H₂O) was added and incubated at 42°C for 30 min. Then 560 ng of anchoroilgo(dT) was added and reaction was incubated at 12°C for 2 h. After 2 h 1 μl of Reverse Transcriptase (PrimeScript RT, Takara) was added and reverse transcription was performed at 42°C for 1 h followed by inactivation at 70°C for 30 min. PCR was performed using a gene specific forward primer (200–600 bp upstream of the poly(A) addition site) and anchor T reverse primers using standard PCR programme. PCR products were visualized by native PAGE (12% gel). PAT-PCR was performed in oocytes collected from n = 3 fish/group. Representative gel images are shown.

H&E staining and histology

Female zebrafish (n = 3/group) were collected three days after the last breeding event and sampled for histopathological examination. Briefly, the fish were fixed in Davidson fixative solution and dehydrated through increasing ethanol series. Then the fish were embedded after immersion in liquid paraffin at 60°C for 2 h. The Formalin-Fixed Paraffin Embedded (FFPE) were sectioned at 3 μm-thickness, and then rehydrated using Xylene and Ethanol solutions. Finally, fish sections were stained with Hematoxylin and Eosin for morphological analysis. Slides were observed using a Zeiss Primovert microscope at 4X, and images were acquired using Axiocam 208 (colour).