HNRNPU's multi‐tasking is essential for proper cortical development

Heterogeneous nuclear ribonucleoprotein U (HNRNPU) is a nuclear protein that plays a crucial role in various biological functions, such as RNA splicing and chromatin organization. HNRNPU/scaffold attachment factor A (SAF‐A) activities are essential for regulating gene expression, DNA replication, genome integrity, and mitotic fidelity. These functions are critical to ensure the robustness of developmental processes, particularly those involved in shaping the human brain. As a result, HNRNPU is associated with various neurodevelopmental disorders (HNRNPU‐related neurodevelopmental disorder, HNRNPU‐NDD) characterized by developmental delay and intellectual disability. Our research demonstrates that the loss of HNRNPU function results in the death of both neural progenitor cells and post‐mitotic neurons, with a higher sensitivity observed in the former. We reported that HNRNPU truncation leads to the dysregulation of gene expression and alternative splicing of genes that converge on several signaling pathways, some of which are likely to be involved in the pathology of HNRNPU‐related NDD.

associated with 1q44 and pointed at critical genes that contribute to the phenotypes. [10,11] ZBTB18 (ZNF238) haploinsufficiency was linked to variable corpus callosum anomalies with incomplete penetrance, and AKT3 was associated with microcephaly. [12][13][14] The identification of 11 patients with a submicroscopic deletion at 1q44 led to the understanding that HNRNPU variants mainly cause epilepsy and intellectual disability. [15] Later, the phenotype of additional patients with 1q43q44 microdeletions [14] and identifying insertion/deletion variants of HNRNPU in a large cohort of patients with epileptic encephalopathies [16] supported this observation.
HNRNPU turned out to be one of a notorious list of 90 genes showing a significant mutation burden in 16 000 screened NDD patients. Most of the mutations found in this screen were predicted to be de novo and likely disruptive. [17] Nevertheless, the discovery of individuals with a disruptive variant, specifically in the HNRNPU gene, was essential for establishing a meaningful genotype-phenotype correlation. [18][19][20] More recently, 17 previously unpublished patients carrying HNRNPU mutations were described, bringing the total number of HNRNPU pathogenic variants to 57. [21] This significant number allows researchers to conclude the nature of these mutations and their resulting phenotypes. Nearly all patients carried de novo mutations. Early onset epilepsy, usually before 2 years of age, was found in all reported individuals; 45% of the reported cases were refractory to treatment. Global developmental delay was common to all patients, and a large majority displayed moderate to severe intellectual disability. Neonatal hypotonia was somewhat less common and reported in approximately half of the cases, and craniofacial dysmorphisms appeared in about half of the patients. A small proportion of individuals with pathological HNRNPU variants displayed behavioral abnormalities, including autistic features, aggression, anxiety, and obsessive-compulsive behaviors. About a third of the patients had cardiac defects composed of septal, atrial, and ventricular septal defects. Fifty-eight percent of individuals with HNRNPU-Related NDD had abnormal brain MRI findings. These included enlarged ventricles (ventriculomegaly) and corpus callosum thinning or agenesis. [21] The possibility that less profound malformation, invisible to MRI, exists in patients cannot be excluded. The reported pathogenic mutations span the entire HNRNPU coding sequence, four of which are recurrent (found in two or more individuals), and two are in predicted splice sites.
None of the deleterious mutations mapped to exons 8 and 13; however, they appear to cluster in exons 2, 3, 9, and 10. [21] Interestingly, HNRNPU is not the only member of the HNRNP family implicated in NDDs (Table 1). Gillentine et al. dubbed the collection of HNRNP that are causative in NDD, NDD-HNRNP and hypothesized that the relatively small number of them (5/33, including HNRNP H1, H2, K, R, and HNRNPU) is merely a partial list and that other family members may have shared molecular pathogenesis. [22] Indeed, this approach led to the identification of new NDD-HNRNP genes, including HNRNP AB, D, F, H3, UL1, UL2, and SYNCRIP, and distinguished them as a defining subgroup of NDD. HNRNPUL2 is one of six new risk genes recently identified in a large cohort of autism cases. Interestingly, ASD individuals carrying variants in HNRNPUL2 had no intellectual disability. [23]

HNRNPU/SAF-A FUNCTIONS-IMPLICATIONS IN BRAIN DEVELOPMENT
A fundamental question in understanding HNRNPU-related NDD etiology is the underlying mechanism of the phenotypic presentation of these malformations. Specifically, which of HNRNPU/SAF-A functions are critical during cortical development. As it turns out, this is not an easy question to address due to the complexity and duration of corticogenesis and the numerous functions of HNRNPU that will be discussed later. HNRNPU/SAF-A is ubiquitous as it is widely expressed across brain regions, CNS structures, and multiple human organs. [24] Cortical development is a highly regulated process during several months of embryonic life. [25] The neural tube initially consists of a monolayer of the neuroepithelium (NE). Neuroepithelial cells self-proliferate by undergoing symmetric divisions. Later, the telencephalon will expand at the rostral part of the tube. The NE will give rise to neural progenitors: radial glia (RG) in the ventricular zone (VZ) and later to outer radial glia (oRG) at the outer subventricular zone (osVZ), the main proliferative zone in the primate brain. Excitatory neurons are born in consecutive waves and migrate radially to form the six-layered cortex in an inside-out order (deeper layers precede the formation of outer layers). Most inhibitory neurons emerge from progenitors at the lateral, medial, and caudal ganglionic eminences (LGE, MGE, and CGE, respectively) and undertake a long tangential migratory path that persists into early postnatal ages. The developmental programs that shape the brain are robust. Nevertheless, they are susceptible to genetic or pathological insults that may result in moderate to severe malformations with potentially profound consequences on the individual's well-being. NDD-HNRNP gene expression pattern was consistent with a role in cortical development, with enriched expression in MGE progenitors, RG, and excitatory neurons.

Remodeling the interphase chromosomes
HNRNPU/SAF-A is found in the inner part of the nuclear matrix, a structural framework that persists after the removal of most chromatin [26,27] and is an important nuclear organizer during interphase (review [28,29] ). The SAP domain of HNRNPU/SAF-A mediates DNA binding and assists in the 3D DNA organization in the nucleus. [27,30] It contributes to the maintenance of the "open" active chromatin state, which is more permissive to transcription and occupies the inner territories of the nucleus ( Figure 3). Indeed, the knockdown of HNRNPU caused compaction of the chromatin. [31] Studies of high-resolution chromatin interactions defined units of chromatin structure, [32] including topologically associated domains (TADs) and lamina-associated domains (LADs). Reduction in HNRNPU's levels increased the coverage of LADs and decreased the strength of TAD boundaries, causing a global shift in nuclear organization. This study also demonstrated that HNRNPU is mainly associated with active TA B L E 1 A list of heterogeneous nuclear ribonucleoproteins in the human genome, indicating the gene symbol, cytogenetic locus, and reported genomic disorders.  [105,106] DD/ID, emotional/behavioral issues, speech delay Facial dysmorphism, progressive growth restriction
These structures are detected in DNA or RNA, and their conformation affects gene expression and epigenetic modifications. [34][35][36][37][38][39][40] HNRNPU/SAF-A's binding to caRNAs and the formation of the nuclear mesh depend on its oligomerization which is driven by repeated ATP binding and hydrolysis cycles. The nuclear matrix consists of additional proteins, including lamins, Matrin-3, and NUMA but also contains an impressive amount of insoluble nuclear RNA. In fact, HNRNPU/SAF-A, unlike other matrix proteins, was released from the matrix following RNAase treatment suggesting a requirement for RNA in determining HNRNPU/SAF-A localization and solubility. [41] HNRNPU/SAF-A promotes inter-chromosomal interaction together with Functional Intergenic Repeating RNA Element (Firre), a nuclear noncoding lincRNA that evades X inactivation. Firre localization bridges at least five cross-chromosomal loci within the nucleus, affecting the high-order nuclear architecture. [42] Nevertheless, HNRNPU/SAF-A has an additional and more direct effect on transcription. It is involved in the transcriptional regulation of specific transcription factors, including Brn4 [43] and Oct4, [44] associated with elements within their promoter regions. Partnering with BRG1 (Brahma Related Gene 1), an ATPase in the nuclear matrix, the SAF-A/BRG1 complex interacts with RNA Polymerase II (Pol II), an event required for Pol II-mediated transcription. [44] Chromatin organization during DNA replication Local DNA decompaction is a prerequisite for the replication progression. [45] Regions rich in DNA-protein complexes with secondary DNA structures and repetitive DNA sequences located at centromeric or telomeric areas pose challenges to the replication process. Failure to replicate such regions may lead to genome instability and cytogenetic lesions. These appear particularly in common fragile sites (CFSs) on metaphase chromosomes when cells encounter replication stress. [46] HNRNPU/SAF-A was critical for robust DNA replication and full replication licensing in G1. Knockdown of the protein led to increased spacing between replication origins, slowing the replication fork progression, creating replication stress, and increasing cellular quiescence. [47] Shortly after exposure to ionizing irradiation and induction of double strands breaks, HNRNPU/SAF-A transiently clusters at damage sites. This localization is regulated by phosphorylation at Ser59 by (DNA-dependent protein kinase) DNA-PK, thus suppressing base excision repair (BER) initiation and allowing the predominant nonhomologous end joining (NHEJ) repair to proceed. [48]

Mitotic chromatin compaction
During mitosis, the chromosomes undergo profound structural changes critical for successful cell division. Our work [49] and others [50,51] described a cell cycle-dependent localization of HNRNPU/SAF-A to chromatin. HNRNPU/SAF-A was co-localized with DAPI throughout the interphase but was removed from chromatin in the M phase, allowing the chromosome to condense. This cellular localization pattern was apparent in coronal sections of the mouse VZ where the dynamic apicobasal movement of RGs nuclei occurs. This phenomenon, known as interkinetic nuclear movement (INM), is the oscillatory movement of the nuclei in correlation with cell cycle progression. [52] The M phase occurs at apical locations, close to the ventricle, while S-phase occurs basally at the proliferative zone's border (VZ/sVZ). During M-phase, HNRNPU/SAF-A is dissociated from the mitotic chromosome, however, it is overall more abundant. [49] HNRNPU/SAF-A binds spindle microtubules in cycling HeLa cells and is later found in kinetochores. [53] This shift in localization of HNRNPU/SAF-A is regulated by phosphorylation by Aurora kinase B at two locations on the SAP domain. [53,54] Failure of HNRNPU/SAF-A to evict the condensed chromosomes from nuclear RNA leads to a high frequency of chromosome misalignment during metaphase and to improper chromosomal segregation during anaphase due to disruption of CENP-E localization and chromokinesin function. [50] The regulation of 3D chromosomal organization is crucial during corticogenesis. It is highly plastic during an extended time window of brain development (review [55] ). Gene expression programs are driving many aspects of neurogenesis with high fidelity. Precise regulation of gene expression patterns dictates the brain's regionalization, proliferation, differentiation of neural progenitors, specification of neuronal subtypes, neuronal migration, and circuit formation, ensuring the development of a properly functioning brain. [56] The progressive loss of NPCs' neurogenic potential during neocortical development is associated with chromatin condensation on a large-scale. [57] During the diversification of cortical cells, changes in chromatin accessibility often precede gene expression suggesting lineage priming exists. [58] The role of HNRNPU/SAF-A in chromatin organization was discussed while Emx1-Cre driving the same double mutations allowed a thin and almost non-proliferative cortex to emerge. [59] Similarly, conditional loss of Brg1 led to microcephaly, cortical thinning, absence of hippocampus, and neuronal migration defects. [60] The Sonic Hedgehog (SHH) pathway is pivotal during neurogenesis and neural patterning. HNRNPU/SAF-A scaffolding activity and its binding to the 5′-UTR of Shh was suggested to affect long-distance regulation of Shh expression via chromatin rearrangement in the limb and may also be relevant to the developing brain. [61] Despite the appreciated importance of chromatin organization and brain development, a direct link to HNRNPU function in this context has not been demonstrated.

X chromosome inactivation
Another chromatin remodeling activity that involves HNRNPU/SAF-A is X chromosome inactivation (XCI). XCl is a process that modi- is tethered close to the nuclear periphery or the nucleolus. Xist (X inactive specific transcript) is a long noncoding RNA (lncRNA) that is necessary and sufficient for XCl. [62,63] Xist recruits silencing factors to form a Xist RNA-protein complex that adopts a modular and evolutionarily conserved architecture [64] and facilitates the re-organization of Xi. [65,62] HNRNPU/SAF-A, with its two nucleic acid binding modalities, SAP and RGG, is required to accumulate Xist RNA on the Xi. [66] HNRNPU/SAF-A participates in tethering the XIST-RNP to the inactive X chromosome while other XIST-associated proteins ensure its binding to the nuclear lamina. Together XIST RNP complex acts as a bridge to bring the X chromosome to the nuclear periphery for remodeling and silencing. [62,67] HNRNPU/SAF-A forms a mesh-like structure within the nucleus, which is dynamic and transcriptionally sensitive. [33] High-resolution images of its nuclear distribution reveal a super-structure of interconnected condensates. [68] These condensates were considered liquid-liquid phase-separated. [28] They may represent a spontaneous ability of HNRNPU/SAF-A to de-mix and form the observed pattern.
The finding supports this idea that HNRNPU/SAF-A was enriched in purified processing bodies (P-bodies). [69] In these cytoplasmic condensates, HNRNPU/SAF-A is one of the other RNA-binding proteins and translation repressors, assisting in separating P-body mRNPs (mRNA molecules coated with RNA-binding proteins) from the other transcripts.
The impact of X inactivation on brain development is not fully understood. Deletion of Xist in the developing mouse brain resulted in reactivation of the X chromosome in a small fraction of neurons and astrocytes and deregulation of histone repressive markers. Still, it had a subtle impact on the brain. [70] In a different study, over-expression of X-linked genes was suggested as a common mechanism for developing psychiatric disorders. [71] HNRNPU, AN RNA-BINDING PROTEIN THAT AFFECTS RNA HNRNPU/SAF-A RNA binding capabilities are impressive, some of which were discussed in the former paragraph in the context of chromatin remodeling. HNRNPU/SAF-A interacts with hundreds of RNAs in the nucleus, its primary location, and the cytoplasm to which it can shuttle. [72] CLIP-seq data suggests that HNRNPU/SAF-A binds to all the classes of regulatory non-coding RNAs. [73] Subsequently, HNRNPU/SAF-A has been implicated in several RNA processing, sorting, and stability aspects.

Regulation of alternative splicing
Alternative splicing (AS) increases the genome information coding capacity and is an essential regulatory mechanism, particularly during neural development. [74] AS is abundant, involving over 90% of human genes. [75] The emergence of specific splice variants of a particular transcript at any given time and location can modulate cellular activities; thus, regulation of the AS is crucial. The splicing event will reflect the collective action of multiple RNA binding proteins (RBPs) occupying specific sequences that interpret the regulatory information of a particular RNA target in a context depended manner. [76] Cis-and trans-elements constitute the "splicing code". [77] In recent years, advances toward cracking the code have been achieved by implementing deep learning algorithms. [78] Among the classic trans-regulatory elements are heterogeneous nuclear ribonucleoproteins (HNRNPs) and SR-rich (serine/argininerich) splicing factors (Figure 4). The entire repertoire of AS regulators is not known; however, genome-wide a CRISPR-Cas9-based screen for splicing regulators revealed that the collection of genes that impact AS is vast. [79] SR proteins have typical arginine and serine residues that form the arginine/serine (RS) domain, one or more RNA-binding domains. In humans, the family consists in human of 12 serine/argininerich splicing factors (SRSF) 1-12 (review [80] ). SR proteins are involved in recruiting the U1 snRNP to the 5ʹ splice site and U2 auxiliary factor (U2AF) to the 3ʹ splice site through protein-protein interactions in the early steps of spliceosome assembly (review [81] ). Generally, SR splice factors are considered positive splicing regulators, promoting exon inclusion by interacting with exonic (ESEs) and intronic (ISEs) splicing enhancers. However, some can bind to intronic splicing silencers and inhibit U1 and U2 recruitment. [81] Similar generalization categorizes hnRNPs as negative regulators of splicing, antagonizing the function of SR splice factors by binding to exonic (ESSs) or intronic splicing silencers (ISSs) sequences. [82] Such an antagonistic effect was demonstrated in the tissue-specific AS of the insulin receptor gene (INSR), in which a mechanism of steric competition between SRSF1 and HNRNPA1 determines the exon recognition by the splicing apparatus. [83] HNRNPC can inhibit or enhance alternative exon inclusion depending upon the position of HNRNPC particles on the nascent transcripts. [84] As mentioned earlier, crosstalk between members of the hnRNP family exists, resulting in a common regulation of specific AS events by multiple hnRNP proteins. [85] A comprehensive database listing components of various purified splicing complexes, supports the idea that HNRNPU/SAF-A is not a core component of the spliceosome. [86] Nevertheless, its identification in a non-bias screen as a significant regulator of SMN2 exon7 splicing led to the finding that HNRNPU/SAF-A binds to intronic regions of a diverse collection of pre-mRNA. HNRNPU/SAF-A had a negative impact on U2 snRNP maturation. HNRNPU/SAF-A depletion caused an enhanced formation of 17S U2 snRNP followed by a cascade of altered splicing events. [73] Differential recruitment of U2 snRNP to competing splice sites was suggested to determine exon inclusion or exclusion.
In vivo, models confirm that HNRNPU/SAF-A has a global effect on splicing. Conditional deletion of exons 4-14 in the mouse Hnrnpu gene that removed 65% of the coding sequence was designed. [87] A heart-specific [87] or telencephalon-specific [49] Cre-drivers were used to create tissue-specific Hnrnpu loss of function. Hnrnpu mutant hearts had over a thousand significant and variable AS events. Of these, specific splice alterations in genes implicated in cardio-myopathy were identified as the underlying cause for heart defects reported in the mutant mice. A similar number of AS dysregulation events (1187) were detected following Hnrnpu truncation in mice cortices. [49] The idea of a common pathology to multiple HNRNPs fits the tendency of family members to complex with other HNRNP proteins ( Figure 2). It may point to their role as AS regulators as the critical function during corticogenesis. Neuronal maturation from the onset of neurogenesis until adulthood is accompanied by specific and temporally controlled AS events in 32% of the brain-expressing genes. [74] Members of the HNRNPs family are involved in AS by forming multivalent HNRNPs assembly that regulate AS in an additive manner. A genome-wide analysis of the auto and cross-regulation of HNRNP proteins HNRNPA1, A2/B1, F, H1, M, and U revealed that most of the HNRNP-dependent AS events are regulated by multiple HNRNP proteins and that often depletion of individual HNRNPU causes similar sets of AS changes. [85] For example, alternative exons in PPIL2 and AFMID transcripts are regulated in an additive manner by HNRNPD and HNRNPAB proteins, whereas alternative exons in CAPN7, C11orf1, and KIF23 transcripts were explicitly regulated by HNRNPA1 and HNRNPA2B1. [88] Direct involvement of HNRNPU function as a splic-ing factor has been suggested for both brain and heart development, as will be discussed later in this review.

mRNA stability
The level of RNA is the dynamic result of transcription, decay, and transport rates that often act as opposing forces. Meta-data-based modeling of the relative predictive contribution of sequences and biochemical features to RNA half-life pointed at HNRNPU/SAF-A as a potent candidate regulator. [89] Experimentally, the stabilization of TNFa, GADD45A, HEXIM1, HOXA2, IER3, NHLH2, and ZFY mRNA by HNRNPU/SAF-A binding to their 3′UTR was reported. [90] Interestingly, a positive feedback regulatory loop was described between c-Myc and HNRNPU in hepatocellular carcinoma. C-Myc was found to transcriptionally regulate HNRNPU while HNRNPU stabilized c-Myc mRNA. [91] RNA stability was not shown to be directly involved in HNRNPU-NDD pathology. Nevertheless, transcriptomics dysregulation documented following HNRNPU loss of function in both mouse and human models may represent a net outcome of transcription and stability.

Sorting of ncRNAs to EVs
Extracellular vesicles (EVs) are a heterogenous collection of membrane-enclosed vesicles released by nearly every eukaryotic cell type. The content of these vesicles is not random and is biologically significant. The tissue of origin determines it and consists of various biologically active molecules: proteins, metabolites, and nucleic acids.
EVs can thus act as messengers between cells and play an essential role in events such as innate and adaptive immune response, [92] pregnancy, [93] cancer, [94] and more. Cargos such as ncRNAs and mRNA can retain their function and act in the recipient cells and may directly impact the levels of specific transcripts.
Interestingly, RNA-binding proteins were identified in large EVs isolated from human endothelial cells, and an abundance of HNRNPU/SAF-A was noted. [95] HNRNPU/SAF-A knockdown increased the miR content in the EV, while the gain of function had the opposite effect. The authors pointed at a specific miR, miR-30c-5p, as one whose export was the most significantly regulated by HNRNPU/SAF-A in a cell type-specific manner. Excessive export increased miR-30c-5p levels in recipient cells and reduced migratory behavior. Computational binding analysis showed a connection between the binding efficiency of the specific miRs to HNRNPU/SAF-A, and a potential sorting signal present on miRs that are selectively loaded onto EVs was identified. [95] A recent study demonstrates the impact of non-cell autonomous signaling via EVs on neuronal specification and migration during corticogenesis. [96] This field is in its infancy and will likely yield many more insights; thus, at this point, a possible involvement of HNRNPU cannot be ruled out.

Models for HNRNPU loss of function
To dissect HNRNPU/SAF-A modalities during corticogenesis and to better understand the etiology of HNRNPU-related NDD, mouse and human-derived models were used. The first published mouse model proved HNRNPU/SAF-A essential for embryonic viability. Embryos homozygous for a hypomorphic mutation exhibited severe growth retardation at E7.5 and died between E9.5 and E11.5. [97] The lethality of homozygous embryos was also noted when a conditional Hnrnpu allele was excised in the female germ line. [87] Mice heterozygous for Hnrnpu loss of function mutation [98] and conditional deletion mutations of Hnrnpu in either the mouse heart [87] or brain [49]  binding domain at the C terminus. [87] Excision of the allele in the heart and production of a detectable truncated protein was seen using a cardiomyocyte-expressing Cre line, muscle creatine kinase-Cre (Ckmm-Cre). While the authors reported a normal development of the heart, they observed a sudden death of mutant animals two weeks after birth due to significant cardiac dysfunction detected shortly after birth and a progressive ventricle chamber dilation. Despite noticeable changes in the expression levels of genes that could be directly linked to heart function, the more significant effect was on alternative gene splicing. Loss of intact HNRNPU expression in the heart led to an increase in skipped cassette events, some of which in genes that have been proposed to be a significant contributing factor to lethal heart failures, such as Bigheart, Ttn, Pstk, Junctin, Camk2d, and Camk2g. Supporting the hypothesis that AS is a major underlying cause of heart failure is the phenotypic similarity to Srsf1 mutant mice. [99] Transcriptomic dysregulation remains a significant factor contributing to the neurological aspect of HNRNPU's loss of function. Ressler and coworkers compared the transcriptome of cortices from two mice models and two human-derived brain organoids. The Hnrnpu +/113DEL mouse line [100] showed moderate 20%-25% HNRNPU levels in the cerebral cortex compared to a telencephalic-driven heterozygous truncation of Hnrnpu. Embryonic and postnatal mice cortices were compared to organoids derived from two isogenic mutants induced pluripotent stem cell (iPSC) lines. The two lines, one carrying a 1 bp duplication and the other 10 bp deletion, had a significant reduction in HNRNPU mRNA levels but only a modest (25%) reduction in protein level. The study pointed to a widespread dysregulation in relevant pathways, including neurogenesis, morphogenesis, and differentiation.
Human organoids, which were grown in culture for 45 days, and embryonic (E13) but not postnatal (P1) cortices showed a similar pattern of transcriptional dysregulation.
Our recent work [49] suggests a convergence of both transcriptomic and AS dysregulation to a significant developmental pathway as contributors to the cortical abnormalities associated with HNRNPUrelated NDD. We drove a telencephalic-specific Hnrnpu truncation using an Emx-1 Cre driver. Unlike the heart-specific mutation of the same allele, we could not detect a truncated protein in the mouse brains, suggesting this is a loss of function mutation.
We observed a unique requirement of HNRNPU for the survival of cultured neural progenitors in vitro and the entire mouse cortex in vivo. The loss of mutated cortical structures was dramatic and progressive, leaving the mice "cortex-less" yet surviving well into postnatal ages. We attributed this to the activation of p53-dependent death programs that were otherwise under check and recorded accumulation of P53 in dying progenitors. Conditional deletion of p53 allowed the cortex to repaper and enabled us to study gene expression and splicing patterns in the mutated brains. We concluded that the observed phenotypes are a combined effect on genes belonging to critical pathways whose expression levels vary and/or are undergoing abnormal AS without HNRNPU. Many of the dysregulated genes were involved in cell viability and cell motility, causing the death of neural progenitors and abnormalities in neuronal migration. The importance of AS in these events was underscored by our ability to reverse the death of neural progenitors and the radial migratory defects by altering the splicing factor SRSF3. We showed that reducing the levels of the splicing factor SRSF3 opposed HNRNPU's loss of function effects and revered the balance between splice variants of an essential regulator of P53 levels,

CONCLUDING REMARKS
The human brain develops during a prolonged gestation period in an orchestrated, tightly regulated, and robust manner. Deviation from its developmental program may result in severe implications on its function. Defining the cause of complex brain malformations is a continuous challenge, especially when a multifunctional gene product such as HNRNPU is involved.

CONFLICT OF INTEREST STATEMENT
Authors declare that they have no competing interests.

DATA AVAILABILITY STATEMENT
Data sharing does not apply to this article as no datasets were generated or analyzed during the current study.