Fragile X syndrome (FXS) is the most common heritable form of cognitive impairment in humans and the leading monogenic cause of intellectual disability and autism spectrum disorder (ASD), affecting approximately 1:4,000 males and 1:11,000 females (Bagni and Oostra, 2013; Hunter et al., 2014). The molecular cause of FXS arises from loss-of-function mutations in the X-chromosome gene, FMR1. Gene silencing is thought to occur early in embryogenesis (~10–11 gestation; Willemsen et al., 2002; Hecht et al., 2017). In nearly all cases, the observed mutation is an expansion of a CGG trinucleotide repeat in the promoter region of the gene (Verkerk et al., 1991). In unaffected individuals, the CGG region is repeated 5 to 54 times. Individuals harboring between 55 and 200 CGG repeats are defined as premutation carriers, and the full mutation state is defined as greater than 200 CGG repeats (Nolin et al., 2019). At this size, hypermethylation of the repeat region leads to the transcriptional silencing of the FMR1 gene and loss of the protein product of FMR1, Fragile X messenger ribonucleoprotein (FMRP). While it is widely accepted that FXS results from the loss or significant reduction in the function of FMRP, knowledge regarding the early molecular mechanisms that link the loss of FMRP to the neuropathology and cognitive impairments that characterize FXS is currently limited (Richter and Zhao, 2021; Bruford, 2022). As a result, mechanism-based disease modifying therapies remain elusive, with few therapeutic agents progressing to phase III clinical trial. Currently, pharmacological treatment for FXS is limited to symptomatic relief of behavioral comorbidities.

In addition to the two major pathogenic allele variants (Full; PM), mitotic instability during neurodevelopment frequently results in somatic cell mosaicism both in CGG repeat size and methylation status of the FMR1 allele. Mosaicism is known to modify phenotypic outcomes in FXS and is linked to a reduction in FXS severity (Hagerman et al., 1994; Loesch et al., 2012; Pretto et al., 2014) and the development of the late onset neurodegenerative disorder, Fragile X tremor ataxia syndrome (Hwang et al., 2016). The heterogenous mechanisms and phenotypes observed in FXS-associated disorders, reinforces the importance of developing model systems that allow insight into the contribution of variable mutant FMR1 alleles to phenotypic outcomes. Given the poor translation of preclinical FXS model systems towards clinically actionable targets to date, the development of tractable and representative models of human neurodevelopmental disorders is crucial to our ability to draw physiologically relevant conclusions about the mechanisms of FXS pathogenesis during early neurodevelopment.

FXS patient-derived post-mortem tissue has provided valuable insights on the impact of FMRP loss in human neurons but cannot recapitulate developmental processes (Hinton et al., 1991; Wisniewski et al., 1991; Irwin et al., 2001). Therefore, FXS patient-derived post-mortem tissue does not provide the ability to study the mechanisms contributing to dysregulated neurogenesis throughout development. Our current understanding of the mechanistic basis of FXS progression during development have, until recently, been largely derived from either FMR1 knock out (The Dutch-Belgian Fragile X Consortium, 1994) or CGG knock in mouse models (Bontekoe et al., 2001). Although an invaluable resource for drug screening and evaluating phenotype development within the in vivo niche, mouse models of FXS do not recapitulate the molecular hallmarks of the FXS mutation as they occur in humans, in particular hyper-methylation of the FMR1 promoter (Brouwer et al., 2007).

It is therefore necessary to develop and utilize neural cell culture models to investigate how the loss of FMRP manifests itself during human neurogenesis. FXS affected human embryonic stem cell (hESC) lines and induced pluripotent stem cells (iPSCs) have provided initial insights into the molecular and cellular changes induced through the loss of FMRP in human neurons (Urbach et al., 2010; Sheridan et al., 2011; Bar-Nur et al., 2012; Doers et al., 2014; Esch et al., 2014; Halevy et al., 2015; Kaufmann et al., 2015). One key finding was that methylation-mediated silencing of the FMR1 gene and the heterochromatinization of the gene promoter were shown to be developmentally dependent (Eiges et al., 2007). The precise developmental timing and mechanism of gene silencing has yet to be fully elucidated, and it is likely that greater variation in the population exists than previously understood. However, evidence suggests that FMR1 is expressed in FXS hESCs, and gene silencing occurs upon differentiation. In contrast, FXS patient-derived hiPSCs exhibit silencing of the FMR1 gene in both the pluripotent and the differentiated states (Urbach et al., 2010) indicating that FXS patient-derived iPSCs cannot recapitulate the developmentally dependent silencing of FMR1. Premature gene silencing brings into question the validity of FXS hiPSC cell lines in recapitulating the effect of FMRP loss during neural development.

Direct cell reprogramming offers an alternative and efficient means to derive a self-renewing and highly expandable source of multipotent human induced neural precursor cells (hiNPC) with the potential to differentiate to mature neurons and glia in vitro while bypassing pluripotency. Many hiNPC reprogramming protocols employ integrative viral vector delivery systems with a concomitant risk of insertional mutagenesis and enhanced genotoxicity. Alternatively, we have demonstrated the ability to directly generate hiNPCs from adult human dermal fibroblasts using chemically modified mRNA (cmRNA) for the neural genes SOX2 and PAX6 (Connor et al., 2018). Furthermore, we recently reported the ability to specifically generate human induced dorsal forebrain precursor cells (hiDFP) using cmRNA SOX2 and PAX6 direct reprogramming with subsequent differentiation to functionally mature glutamatergic neurons (Edwards et al., 2021). We will utilize our direct reprogramming technology to generate, for the first time, hiDFPs from FXS patient-derived fibroblasts. We propose that utilization of a direct-to-hiDFP protocol to study FXS will provide a more robust mechanism than the current use of hiPSCs by negating concerns regarding the premature inactivation of the FMR1 gene as seen in FXS patient-derived hiPSC lines (Bhattacharyya and Zhao, 2015). This study represents a novel approach to develop a robust and efficient human-based cell model of FXS.

In FXS, the size and stability of the highly polymorphic CGG repeat tract in FMR1 correlates with distinct phenotypic outcomes and severity. Carriers of the pre-mutation allele (55–200 CGG) for example, exhibit distinct phenotypes compared to full FXS (>200 CGG) affected individuals (Hagerman and Hagerman, 2004). Additionally, repeat size mosaicism within an individual is also linked to moderately increased levels of FMRP compared to full mutation alleles in males, and milder cognitive phenotypes (Baker et al., 2019). While increased repeat length confers susceptibility and elevated risk of FXS development, the methylation characteristics of the FMR1 allele are highly predictive of FXS presentation. Methylation mosaicism for example has been linked to improved clinical outcomes in FXS (Pretto et al., 2014), while unmethylated full mutation alleles (UFM) in male carriers exhibit a reduction in FXS phenotype severity (Hagerman et al., 1994; Loesch et al., 2012) but are susceptible to the late onset-neurodegenerative disorder FXTAS (Hwang et al., 2016). The development of modelling strategies that recapitulate the diverse characteristics of the FMR1 locus as it occurs in humans are required to better understand the relationship between genotype, epigenotype and phenotype throughout neuronal development. Such model systems will better facilitate our understanding of the mechanisms, timing, and impact of gene silencing throughout development.

This study represents the first reported derivation of FXS-affected cortical neurons following cmRNA-based direct reprogramming of patient fibroblasts to dorsal forebrain precursors. Methylation of the promoter is a key molecular hallmark of FXS in human tissue and an important modifier of phenotype severity. These effects are mediated by methylation mediated silencing of the FMR1 gene and subsequently loss of FMRP, an important regulatory protein in neurodevelopment. Preservation of the methylation signature in the 5’upstream FMR1 promoter in our study indicates that SOX2/PAX6 cmRNA-mediated reprogramming and cortical differentiation did not induce epigenetic instability in either control or FXS cell lines. These findings also lend support to the hypothesis that correlations between the epigenotype and phenotype identified are representative of FXS, rather than artefacts of an unstable methylation signature introduced by the reprogramming mechanism or culture conditions. Interestingly, we identified one FXS cell line that exhibited FMR1 expression and a promoter methylation signature consistent with a mosaic status. While unable to undertake statistical analysis, the mosaic FXS+ line provided insight into the effect of partial FMRP expression on neurogenesis. In summary, transcriptional (FMR1), epigenetic (% methylation), and phenotypic (FMRP) characterization of FXS hiDFPs and derived neurons confirmed that our model recapitulates the molecular hallmarks of FXS.

To confirm whether our SOX2/PAX6 cmRNA direct reprogramming strategy could promote the generation of FXS-affected dorsal forebrain precursor cells, the expression of relevant lineage markers were assessed. Increased expression of genes required for specification of a dorsal telencephalic precursor fate were identified in both control and FXS- hiDFPs, including FOXG1, NGN2, TBR2, TBR1 and CTIP2. This is also consistent with decreased expression of the ventral forebrain marker DLX2. The pan-precursor marker Nestin was also highly expressed across all hiDFPs cohorts. These findings were reinforced by protein expression analysis confirming expression of the phenotypic markers NGN2, FOXG1, TBR2, indicating dorsal forebrain precursor development. While no significant difference in the expression of the proliferative marker Ki67, or neural precursor markers, NGN2, TBR2 were detected, a significant increase in FOXG1 expression was identified in FXS- hiDFPs. FOXG1 is an important regulator of cell cycle progression and is an established ASD candidate gene (Mariani et al., 2015). Differential expression of this key developmental regulator may be indicative of dysregulated cell cycle progression of FXS- hiDFPs towards a neuronal lineage.

Following reprogramming, we wanted to confirm whether an enriched population of glutamatergic neurons could be generated from SOX2/PAX6 cmRNA derived dorsal forebrain precursors. Additionally, differential expression of genes regulating neuronal fate specification, neuronal differentiation, cytoskeletal and axonal guidance, and synaptic signalling have been reported in FXS affected neurons (Halevy et al., 2015; Ferron, 2016; Lu et al., 2016; Danesi et al., 2018; Utami et al., 2020; Raj et al., 2021). Therefore, the current study investigated whether differential expression of key neuronal and synaptic genes could be identified in FXS compared to control hiDFP-derived neurons. Gene expression analysis following 14 days of differentiation indicated that FXS- hiDFP-derived neurons exhibited hallmarks of impaired neuronal development and maturation with a low fold-change increase of immature (TUBB3) and mature (MAP2) neuronal markers. Similarly, decreased expression of mature neurons markers NSE and MAP-T were observed. In FXS+ hiDFP-derived neurons, expression of mature neuronal markers was either low or decreased, suggesting partial FMRP expression was insufficient to mitigate impaired maturation at the transcriptional level.

Cortical neuron development was confirmed by increased expression of cortical layer specific neuronal markers in control and FXS hiDFPs following differentiation. Increased expression of the layer II-III specific markers BRN2 and CUX1 in control hiDFP-derived neurons is indicative of upper layer neuron development (McEvilly et al., 2002; Sugitani et al., 2002; Bulchand et al., 2003; Espuny-Camacho et al., 2013). Additionally, expression of TBR1 (layer VI) and CTIP2 (layer V) also indicate the presence of deep layer neurons (Hevner et al., 2001; Espuny-Camacho et al., 2013). Elevated expression of BRN2 was similarly detected in FXS- hiDFP-derived neurons, although no change in CUX1 expression relative to hiDFPs was detected. Although patterns of cortical lamination could not be assessed in our 2D in vitro system, impaired CUX1 expression in layers II/III of FMR1 KO mice have been linked to impaired migration of early born neurons to superficial layers of the cortex during corticogenesis (Lee et al., 2019). Impaired CUX1 expression in FXS- hiDFP-derived neurons may therefore indicate impaired migration and neural positioning competence. Additionally, CUX1 expression has been shown to regulate dendrite morphology and dendritic spine maturation in upper layer cortical neurons in CUX1 null mice (Cubelos et al., 2010). However, these findings have yet to be replicated in human derived FXS neurons. The critical role of CUX1 in dendrite branching and synapse formation may also indicate impaired competence for proper dendrite elaboration and synapse formation in CUX1 deficient FXS- hiDFP-derived neurons. In addition to late born upper layer neuron markers, an increase in TBR1 expression and lack of CTIP2 expression suggests deep layer neuron specification in FXS- hiDFP-derived neurons. The expression of cortical layer markers in FXS+ hiDFP-derived neurons were comparable to controls, with the exception of TBR1, which was highly upregulated. Together, these results suggest a mixed population of deep and upper layer neurons of a glutamatergic lineage in both control and FXS hiDFP-derived neurons, with evidence of impaired neuronal maturation in FXS- but not FXS+ neurons. This contrasts with the findings of Kang et al. (2021), in which a bias towards deep layer neuron specification in FXS iPSC derived neurons was observed. Interestingly, FXS iPSC- derived deep layer neurons exhibited accelerated maturation, increased synapse formation and network hyperexcitability (Kang et al., 2021), contrary to our own findings.

Impaired cognitive function in FXS patients is associated with an imbalance of excitatory and inhibitory synapses (Gibson et al., 2008; Nelson and Valakh, 2015). Efforts to understand the mechanisms underpinning the imbalance of excitation and inhibition during early neural circuit formation in neurodevelopmental disorders, including FXS, are ongoing (Bagni and Zukin, 2019). Impaired synaptic homeostasis is linked to the loss of the activity-dependent functions of FMRP during postnatal periods of synaptic refinement (Tessier and Broadie, 2008; Pfeiffer et al., 2010). Therefore, the signalling events mediating cortical hyperexcitability and impaired inhibitory activity in FXS are thought to be largely mediated by post-transcriptional events and may not necessarily be reflected at the transcriptional level (Antar et al., 2006; Bassell and Warren, 2008). Nonetheless, elevated expression of the SLC17A7 gene during neuronal development is prerequisite for the molecular assembly of functional excitatory synapses (Wojcik et al., 2004), and its upregulation in control hiDFP-derived neurons is indicative of glutamatergic lineage acquisition (Takamori et al., 2000). Although the expression of SLC17A7 was confirmed in FXS- and FXS+ hiDFP-derived neurons, this was greatly reduced compared to controls. We also investigated expression of the synaptic protein encoding genes, SYN1 (synapsin-1), and DLG4 (PSD-95), which have established roles in regulating excitatory glutamatergic synapse function together with SLC17A7 (vGLUT1). While both SYN1 and DLG4 were expressed in control neurons, neither gene exceeded a 2-fold change threshold for expression in FXS- hiDFP-derived neurons following 14 days of differentiation.

FMRP modifies L-type voltage-gated calcium channels and sodium-/calcium-activated potassium channels through regulating mRNA expression at the plasma membrane or binding ion channels directly (Brown et al., 2001; Darnell et al., 2011; Deng et al., 2013; Ferron et al., 2014) CACNA1C encodes the alpha-1 subunit of L-type voltage-dependent calcium channels (Ca𝙑.2,) and is predominantly expressed in glutamatergic neurons of the forebrain. Ca𝙑.2 has been shown to play a central role in activating calcium-mediated signalling pathways necessary for the differentiation of cortical neurons, dendritic growth, and plasticity, and is a known candidate susceptibility gene for a range of related disorders characterized by impaired cognitive function (Paşca et al., 2011). Similarly, hyperpolarization-activated cyclic nucleotide-gated channels (HCN) play fundamental roles in regulating neuronal excitability, and impaired homeostatic H-channel plasticity has been observed in FMR1 deficient mice (Brager et al., 2012). While the expression of CACNA1C was upregulated in both control and FXS hiDFP-derived neurons, fold change expression was lower in FXS- hiDFP-derived neurons. Additionally, while HCN2 was expressed in controls, fold-change expression remained low in FXS- hiDFP-derived neurons. Low CACNA1C expression coupled with a lack of HCN2 expression above a 2-fold threshold in FXS- hiDFP-derived neurons may indicate impaired development of functional synapses compared to controls. Further studies examining whether alterations in the wider synaptic milieu exist at the protein level are required. The functional consequences of dysregulated protein expression at the synapse would provide clarity on the impact of the reduced gene expression of key synaptic markers identified in our study.

Following 14 days of differentiation, we detected a significant reduction in the yield of FXS- hiDFP-derived neurons compared to controls. This is consistent with previous reports demonstrating that FMRP deficiency during neural precursor development leads to reduced neuronal production and maturation (Luo et al., 2010; Guo et al., 2011). However, other studies investigating neurogenesis in FXS human pluripotent stem cells have reported both impaired (Boland et al., 2017), enhanced (Castren et al., 2005) and no change (Zhang et al., 2018) in neuronal yield. Despite the decrease in the neuronal yield of FXS- hiDFPs following differentiation, neuronal subtype specification to a glutamatergic neuronal fate was not affected. Interestingly, while no difference in the proportion of neuronal cells expressing a glutamatergic phenotype was observed, we detected a significant increase in the percentage of S100β + cells in FXS- hiDFP-derived cultures compared to controls following 14 days of differentiation. An increase in the proportion of S100β + cells following differentiation of FXS hiDFPs was surprising given the enhanced expression of FOXG1 observed following reprogramming. FOXG1 has been shown to drive a developmental bias towards neuronal over glial progenitor development (Brancaccio et al., 2010), as well as promote neurogenic competence in mouse derived astrocytes following forced expression (Ma et al., 2018). However, the mechanism by which elevated FOXG1 might affect this neurogenic bias in the context of FMRP loss has not been reported. Reductions in neuronal yield and maturation in FXS studies have been attributed to impaired neurogenesis due to a shift in fate specification towards a glial cell lineage (Luo et al., 2010; Sunamura et al., 2018; Brighi et al., 2021). The mechanisms by which FMRP loss during neurogenesis may alter the ratio of neurons to astrocytes, and when this occurs remains unclear. However, these studies support the importance of FMRP in regulating neuronal lineage specification as well as survival and maturation.

Neuronal function and network development are hallmarks of proper neuron development. We therefore sought to confirm the generation of functional neurons and to assess differential functional properties in FXS- hiDFP derived neurons compared to controls. Elevations in intracellular calcium are an indirect result of action potential-driven membrane depolarization. We therefore utilized calcium imaging as an unbiased population level proxy for neuronal activity. Live-cell calcium imaging in control hiDFP-derived neurons confirmed intracellular calcium responses consistent with functional glutamatergic receptor expression. Notably, FXS- hiDFP-derived neurons exhibited dampened calcium responses following glutamate stimulation across a range of concentrations compared to controls. While impaired functional maturation is consistent with impaired development of the dendritic arbor and synaptic protein expression, our findings contrast with reports of increased calcium conductance and hyperexcitability associated with FMRP loss in both rodent and human cell models of FXS (Contractor et al., 2015; Achuta et al., 2018; Danesi et al., 2018; Brighi et al., 2021). Follow-up studies are therefore required to determine the mechanism of the impaired glutamate-induced calcium response observed in FXS- hiDFP-derived cortical glutamatergic neurons.

Alterations in the local repertoire and abundance of synaptic proteins in the postsynaptic compartment are linked to modification of the activity-dependent intracellular calcium response mediated by glutamatergic receptors. Expression of the scaffolding protein PSD-95, for example, modifies glutamate transmission through alterations in NMDA and AMPA-receptor recruitment, stability, and function in the postsynaptic compartment (El-Husseini et al., 2000; Ehrlich and Malinow, 2004; Chen et al., 2015), and is associated with cognitive phenotypes in a range of neurodevelopmental disorders (Tsai et al., 2012; Lelieveld et al., 2016). Determining whether the reduced PSD-95 protein abundance observed in our model is associated with altered subunit specific NMDA and AMPA receptor localization and function may clarify the functional impact of impaired synaptic protein expression on calcium flux in our model.

FMRP has established functions in the nuclear export of putative FMRP mRNA targets, regulating localized activity-dependent translation of proteins important for the integrity of cytoskeletal architecture and synaptic maturation. Given proper elaboration of the dendritic arbor is required for neuronal network formation and function (Lefebvre et al., 2015), the morphological properties of FXS hiDFP-derived neurons were assessed. In addition to a reduction in neuronal yield, morphological abnormalities in FXS- hiDFP-derived neurons consistent with impaired structural maturation were identified. Morphological analysis revealed a reduction in dendritic outgrowth and complexity as reflected in a reduction in total and primary dendrite lengths, average number of branches, and branch points in FXS- affected neurons. Interestingly, this difference was not observed relative to controls during early neurogenesis at day 7, suggesting that while FXS- hiDFP-derived neurons fail to acquire a properly ramified dendritic arbor consistent with mature neuron development, initial dendritic outgrowth remains unaffected. The absence of morphological deficits in immature FXS- hiDFP-derived neurons at day 7 of differentiation, followed by delayed morphological development by day 14 suggest defective neuronal maturation not impaired differentiation of hiDFP to neurons. The reduction in neurite length and dendritic arborization identified in FXS- hiDFP-derived neurons is consistent with phenotypes observed in both neonatal and adult FMR1 KO or knock-down mouse models of FXS (Castren et al., 2005; Shen et al., 2019; Yau et al., 2019; Lannom et al., 2021) and human FXS patient derived iPSCs (Sheridan et al., 2011; Doers et al., 2014; Achuta et al., 2018; Utami et al., 2020). Additionally, neurospheres derived from human fetal postmortem tissue and human iPSC-derived neurons have demonstrated altered differentiation characterized by fewer and reduced neurite extensions (Castren et al., 2005; Doers et al., 2014). These findings however, conflict with a later report by Boland et al. (2017) in which an increase in the neurite length of FMR1 deficient iPSC-derived neocortical glutamatergic neurons was observed at day 12 of differentiation. Interestingly, Boland et al. (2017) observed the effect of FMR1 loss on neurite morphology was transient, with no net difference in phenotype observed by day 15 of differentiation relative to controls. Morphological assessments in the study by Boland and colleagues were limited to early neurogenesis. However, downregulated expression of key genes regulating axonal guidance and neurite growth at day 80 suggest impaired neuronal and synaptic maturation may also occur during later stages of neuronal development (Boland et al., 2017).

In addition to the cell intrinsic mechanisms effected by neuronal FMRP loss, the morphological abnormalities identified in FXS- hiDFP-derived neurons may be attributable to an increased population of S100β + astrocytes. Jacobs et al. (2010) demonstrated that hippocampal neurons from wild type mice exhibited delayed dendritic arborization and impaired synaptic maturation following co-culture with FMR1−/− astrocytes. Additionally, aberrant dendritic morphologies and reduced pre- and post-synaptic proteins were restored in hippocampal neurons from FMR1 KO mice following coculture with WT astrocytes (Jacobs et al., 2010). More recent studies have indicated that alterations in astrocyte secreted signals such as Thrombospodin-1 (TSP-1; Cheng et al., 2016), as well as aberrant BMP signaling (Caldwell et al., 2022) in FMR1-deficent astrocytes results in impaired neurite outgrowth.

Proper elaboration and maturation of dendritic morphology is linked to synaptic protein expression (Vessey and Karra, 2007) and is a key determinant of synaptic function (Wong and Ghosh, 2002). Therefore, the expression of synaptic proteins required for maturation of the glutamatergic synapse in hiDFP-derived neurons was examined. Furthermore, FMRP regulates the stability and localized translation of putative mRNA targets encoding synaptic proteins necessary for dendritic spine maturation and synaptic plasticity (Cruz-Martín et al., 2010; Pfeiffer et al., 2010; Darnell et al., 2011; Deng and Klyachko, 2021). FMRP loss is therefore implicated in the aberrant synaptic signaling and network hyperexcitability thought to underpin impaired cognitive function in FXS (Contractor et al., 2015). The current study therefore sought to determine whether key synaptic proteins were differentially expressed in FXS- hiDFP derived neurons. The expression of Synapsin-1, PSD95 and vGLUT1 positive puncta were indicative of pre- and postsynaptic glutamatergic specialization in both control and FXS hiDFP-derived neurons. However, the density of PSD95 and vGLUT1 puncta was significantly lower in FXS- neurons following 14 days of differentiation compared to controls. PSD95 is a key synaptic scaffolding protein that regulates the molecular organization of the protein enriched region of the postsynaptic membrane of excitatory neurons (Sheng and Kim, 2011). It has also been shown to regulate the developmentally appropriate pruning of dendritic spines and thus synaptic strength throughout development (Nikonenko et al., 2008; Zhang and Lisman, 2012; Cane et al., 2014). The observation that PSD95 expression was lower in FXS- hiDFP-derived neurons was surprising given the canonical role of FMRP in inhibiting the translation of its putative mRNA targets. Specifically, FMRP loss has been linked to enhanced association of target mRNA with translating polyribosomes (Muddashetty et al., 2007) and impaired degradation of PSD95 at the synapse (Tsai et al., 2012). In contrast, a significant decrease in PSD95 mRNA and protein expression due to loss of FMRP-mediated mRNA stability was detected by Zalfa et al. (2007) in FMR1 KO mice. Interestingly, differential expression of PSD95 was only detected in the hippocampus and cerebellum and not cortical neurons of FMR1 KO mice. This is consistent with the findings of Todd et al. (2003) who observed only a transient increase in PSD95 expression in the cortical neurons of FMR1 KO mice compared to WT (Todd et al., 2003).

As changes in vGLUT1 expression are known to modify the strength of excitatory synaptic signaling (Daniels, 2004; Zhang et al., 2019) the observed reduction in the density of vGLUT1 puncta may suggest impaired excitatory synapse development in FXS- neurons compared to control neurons. However, reduced expression of vGLUT1 in FXS- neurons was unexpected given the reported role of FMRP in negatively regulating the translation of synaptic mRNA and reducing synapse number to regulate network activity (Pfeiffer and Huber, 2007). This is supported by a recent report by Brighi and colleagues who observed an increase in the co-localization of vGLUT1+/PSD95+ puncta during early development in FXS hiPSCs with a concomitant increase in spontaneous network activity (Brighi et al., 2021). Interestingly, this phenotype was found to be transient. Similarly, an aberrant increase in synaptic development as determined by an increase in the density of Synapsin-1+/PSD95+ synaptic boutons and enhanced neuronal excitability was detected in FXS hiPSCs (Kang et al., 2021). Interestingly, evidence of increased or unchanged expression of vGLUT1 containing synapses in the neocortex of FMR1 KO mice was shown to be dependent on layer specific neuronal identity with no change observed in cortical neurons (Wang et al., 2014). Collectively, the reduced density of PSD95 and vGLUT1 puncta may be indicative of impaired or delayed development of mature functional synapses in FXS- hiDFP-derived neurons. This is consistent with the observed impairment in expression of mature neuronal markers and evidence of immature dendritic arbor development. However, validating and quantifying the co-localization of postsynaptic (PSD95) and presynaptic (Synapsin-1, vGLUT1) proteins as a representative feature of functional synapses within FXS- hiDFP-derived neurons is necessary to confirm whether loss of protein expression correlates with a loss of synapse density in our study.

Although differential expression of key markers of neuronal maturation and function were identified in our model system, low sample size is a key limitation of our study. The number of independent biological replicates reported in cell reprogramming studies is typically low, reinforcing the practical limitations of scaling such strategies to align with the large sample sizes often required to power such analyses (Germain and Testa, 2017). Multiple clones derived from 1 to 2 individual donors are frequently utilized in iPSC-based studies to improve sample size. This approach often artificially reduces group mean variability when multiple clones from the same individual are treated as statistically independent (Germain and Testa, 2017). Our direct reprogramming method is not susceptible to the genomic variability that arises through clonal expansion (Abyzov, 2012; Young et al., 2012) and the extended culture times (Mayshar et al., 2010) observed in iPSC-based approaches. The donor-to-donor variability identified in this study is thought to reflect the true heterogeneity of independent biological replicates included. Although scalability is limited by the availability of relevant FXS affected donor tissue, increased sample sizes are required to validate our findings in future studies and to capture smaller effect sizes.

In conclusion, this study represents the first reported derivation of FXS-affected cortical glutamatergic neurons following reprogramming of patient fibroblasts directly to dorsal forebrain precursors. Importantly, the FXS-associated phenotypes detected in this study were identified in human neuronal cells that recapitulate the molecular hallmarks of the FXS mutation as it occurs in FXS-affected individuals. Differential expression of mature neuronal markers suggests impaired neuronal development and maturation in FXS- hiDFPs derived neurons compared to controls. This is supported by the reduced yield of FXS- hiDFP-derived neurons and increase in FXS- affected astrocytes following 14 days of differentiation. Additionally, FXS- hiDFP-derived cortical neurons exhibited dendritic growth and arborization deficits characterized by reduced neurite length and reduced neurite branching consistent with impaired neuronal maturation. The significant decrease in the density of pre- and post- synaptic proteins in FXS- hiDFP-derived neurons suggests impaired excitatory synapse development. While consistent with an immature neuronal phenotype, further co-localization and functional maturation analysis is required to determine whether this finding is consistent with aberrant synapse formation and network function reported in both FMR1 KO animal and patient derived FXS-iPSCs. The differential phenotypic outcomes reported in this study and others are likely to result from differences in the developmental stage and relative maturity of neurons derived by various modelling strategies, and likely indicate regional and neuronal cell subtype specific effects of FMRP loss, which are often poorly defined. Whereas the aberrant morphological phenotypes observed in human postmortem tissue captures the accumulated impact of FMRP loss, both established FMR1 KO mouse models and emerging cell reprogramming based strategies indicate the importance of the stage dependent effects of FMRP loss on phenotypic outcomes. The functional effects of transient morphological phenotypes to the neurodevelopmental progression of FXS reinforces the importance of defining the relative maturity of derived neuronal populations to allow meaningful comparisons between studies, as well as the need to refine the temporal resolution of developmental modelling strategies to better capture and understand the effects of intermittent phenotypes throughout development. Despite discrepancies in phenotypic outcomes reported, collectively these studies indicate that FMRP is strongly implicated in the morphogenesis of the dendritic arbor and synaptic maturation. The magnitude and direction of these effects appear to be contingent on developmental stage, neuronal subtype and modelling strategy employed.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

Ethical approval was not required for the studies on humans in accordance with the local legislation and institutional requirements because only commercially available established cell lines were used.

BC: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Validation, Writing – original draft, Writing – review & editing. NE: Formal analysis, Investigation, Methodology, Resources, Writing – original draft, Writing – review & editing. CC: Formal analysis, Investigation, Methodology, Validation, Writing – review & editing. AM-C: Investigation, Methodology, Writing – review & editing.

This work was supported by the Neurological Foundation of New Zealand and the Kate Edger Educational Charitable Trust (KEECT).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.