Verheij syndrome (OMIM #615583) involves a spectrum of phenotypes, including neurodevelopmental delay, intellectual disability, brain malformations, microcephaly, short stature, and ocular, craniofacial, skeletal, cardiac, and renal anomalies (Fennell et al., 2022). This syndrome is caused by a deletion in the 8q24.3 region, where the PUF60 gene is located (Verheij et al., 2009; Miao et al., 2024). PUF60 is involved in 3′ splice-site recognition, interacting with U2AF in RNA binding and splicing activation (Hoogenboom et al., 2024; Hastings et al., 2007). The majority of cases of Verheij syndrome are due to de novo mutations; however, rare cases show an autosomal dominant inheritance pattern (Verheij et al., 2009; Sivasubramanian and Ayyavoo, 2024). Haploinsufficiency of PUF60, due to deletions spanning from 78 kb to 1 Mb, has been found to be the driver of Verheij syndrome, with the copy number variants (CNVs) affecting the dose of multiple genes depending on the size of the deletion (Hoogenboom et al., 2024; Dauber et al., 2013).

Mutations in EFTUD2, part of the U5 snRNP of the spliceosome, have been identified as the cause of mandibulofacial dysostosis, Guion-Almeida type (MFDGA; OMIM #610536) (Lines et al., 2012; Beauchamp and Jerome-Majewska, 2024). MFDGA is characterized by craniofacial malformations, microcephaly, developmental delay, and dysmorphic appearance but may also include choanal atresia, sensorineural hearing loss, and cleft palate (Guion-Almeida et al., 2006; Wieczorek et al., 2009). The frequency of MFDGA is unknown, with approximately 100 affected individuals identified so far, harboring 86 distinct EFTUD2 mutations (Beauchamp et al., 2020). Most mutations are stop-gain and splicing mutations, with roughly 75% of patients harboring de novo mutations, whereas dominant inheritance is observed in the remaining patients (Huang et al., 2016).

Nager syndrome (OMIM #154400) is a type of acrofacial dysostosis characterized by midface retrusion, micrognathia, absence of thumbs, and radial hypoplasia (Bernier et al., 2012; Czeschik et al., 2013; Petit et al., 2014). Sixty percent of patients with Nager syndrome have mutations in SF3B4, with haploinsufficiency of SF3B4 being the underlying cause of the disorder (Bernier et al., 2012). Rodriguez syndrome (OMIM #201170) is also caused by mutations in SF3B4 (Drivas et al., 2019). The patients have similar features as Nager syndrome patients; however, the phenotype is typically more severe and involves lower limb and cardiac defects (Rodríguez et al., 1990). SF3B4 encodes for SAP49 and is part of the U2 and U12 snRNPs (Will and Lührmann, 2011), functioning in 3′ branchpoint sequence recognition. The frequency of Nager syndrome is unknown, with approximately 100 cases found worldwide.

Mutations in SNRPB cause cerebro-costo-mandibular syndrome (CCMS; OMIM #117650), which is a disorder that includes micrognathia, glossoptosis, cleft palate, and posterior rib gaps (Lynch et al., 2014). SNRPB is part of the Sm ring that is the scaffold for snRNPs in the U1, U2, U4, and U5 subunits (Schwer et al., 2016; Zahoor et al., 2024). Mutations tend to be heterozygous regulatory mutations, with high frequency of mutation in the premature termination codon-containing exons. CCMS is a rare disease, with approximately 80 reported cases (Bacrot et al., 2015).

Mutations in gene SNRPA (OMIM #182285) lead to a yet unnamed syndrome that includes intellectual disability, short stature, and minor craniofacial and hand anomalies (Rangel-Sosa et al., 2018). The mutations are homozygous missense variants in SNRPA, which encodes for an snRNP in the U1 subunit of the spliceosome (Nelissen et al., 1991). Mutations tend to be localized to the first 10–89 amino acids, which is the domain associated with RNA binding (Rangel-Sosa et al., 2018; Jessen et al., 1991).

Richieri-Costa–Pereira syndrome (RCPS; OMIM #268305) is a type of acrofacial dysostosis, in which patients exhibit craniofacial malformations with microcephaly and limb defects (Bertola et al., 2018; Favaro et al., 2014; Hsia et al., 2018). This disorder is due to decreased expression levels of EIF4A3, a member of the exon junction complex in the spliceosome (Le Hir et al., 2016). The decreased levels of this gene are attributed to increased repeats in the 5′UTR of the gene. RCPS is a rare disorder, with less than 50 published cases (Pardo et al., 2021), with a possible phenotypic spectrum related to the number of repeats in the gene (Bertola et al., 2018).

Mutations in TXNL4A cause a disorder known as Burn–McKeown syndrome (BMKS; OMIM #608572). This condition is characterized by choanal atresia, hearing loss, cleft lip/palate, and other craniofacial anomalies (Wieczorek et al., 2014). TXNL4A encodes a component of the spliceosome U5 snRNP, and the mutations in patients lead to reduced expression and ultimately reduced assembly of the snRNP complex. This disease has been found in 20 individuals with biallelic pathogenic variants (Lüdecke and Wieczorek, 2022). Most patients have a loss-of-function deletion in the promoter region of the gene; however, patients have been identified with intronic deletions (Wood et al., 2022).

Mutations in RBM8A, a member of the exon junction complex, resulting in haploinsufficiency lead to a disorder characterized by microcephaly, facial gestalt, cleft lip/palate, and skeletal anomalies (OMIM #274000) (Gamba et al., 2016; Mao et al., 2015). These mutations also involve microdeletions of the 1q21.1 chromosome, resulting in variable syndromic phenotypes (Upadhyai et al., 2020). Occurrence of CNVs in 1q21.1 is rare, with less than 40 reports in the literature (Brunetti-Pierri, et al., 2008).

Au–Kline syndrome (OMIM #616580) is a developmental disorder characterized by hypotonia, intellectual disability, and typical facial features (Au et al., 2019). This disorder is due to variants in HNRNPK, which is part of the spliceosome C complex, resulting in impairment of Hox gene expression (Duijkers et al., 2019). Loss-of-function mutations are also associated with a specific DNA methylation signature (Choufani et al., 2022). All currently known patients have de novo mutations, with some including missense variants and others being deletions of 9q21.32, encompassing HNRNPK (Au et al., 2018).

Mutations in RBM10 cause an X-linked form of cleft palate known as TARP syndrome (Talipes equinovarus, Atrial septal defect, Robin sequence, and Persistent left superior vena cava; OMIM #311900) (Johnston et al., 2010; Gripp et al., 2011). RBM10 is an RNA-binding protein that plays a role in the A complex of the spliceosome, regulating alternative splicing. TARP syndrome is a very rare disorder, with approximately 30 cases reported (Omorodion et al., 2023). Mutations tend to be loss-of-function, occurring either de novo or via X-linked dominant inheritance, in which male children are affected and mothers may present some mosaicism (Johnston et al., 2013).

Craniofacial microsomia (OMIM #164210) is a disorder that includes auricular malformations and underdevelopment of the mandible but may also affect the middle ear ossicles, temporal bone, zygoma, and cranial nerves (Beleza-Meireles et al., 2014; Keogh et al., 2007; Timberlake et al., 2021). The most prevalent genetic cause of craniofacial microsomia is haploinsufficiency of SF3B2, a component of the U2 small nuclear ribonucleoprotein complex. Loss-of-function mutations in SF3B2 account for 3% of sporadic cases and 25% of familial cases, with mutations spread across the entirety of the gene (Timberlake et al., 2021). Craniofacial microsomia occurs in between 1 in 5,600 and 1 in 26,550 births, but mild cases are often difficult to diagnose (Gougoutas et al., 2007).

Taybi–Linder syndrome (TALS; OMIM #210710), or microcephalic osteodysplastic primordial dwarfism type I (MOPD1), is characterized by severe growth delay, microcephaly, intellectual deficiency, bone abnormalities, and other factors ultimately resulting in early mortality (Hagiwara et al., 2021). Roifman syndrome (OMIM #616651) is a disorder characterized by growth retardation, cognitive delay, and spondyloepiphyseal dysplasia (Merico et al., 2015). Lowry–Wood syndrome (OMIM #226960) is a similar disorder characterized by multiple epiphyseal dysplasia, microcephaly, and intellectual disability (Farach et al., 2018). All three of these disorders are attributed to mutations in RNU4ATAC, a small nuclear RNA essential for minor intron splicing (Edery et al., 2011).

The CWC27-related spliceosomopathy (OMIM #250410) is also known as retinitis pigmentosa with or without skeletal anomalies. Although this is mainly categorized as a retina disorder, when the patients have skeletal anomalies, they include craniofacial abnormalities, classifying this disorder also as a craniofacial spliceosomopathy (Xu et al., 2017). Variants of CWC27 are diverse and may result from missense mutations, nonsense mutations, splice-site variants, small insertions, small deletions, and gross deletions (Li et al., 2024).

Craniofacial spliceosomopathies represent a broad array of phenotypes affecting many parts of the body; however, they all have one common denominator: defects in the neural crest-derived craniofacial skeleton, and can be categorized as facial dysostoses (Figure 1). Although the splicing factors affected under these conditions are found across the spliceosome and carry distinct functions (Griffin and Saint-Jeannet, 2020), they all cause a similar craniofacial phenotype, pointing at a possible common root cause and driving the need to further investigate the underlying mechanisms of these disorders.

Mouse models are a hallmark of disease modeling and can be a powerful system for understanding disease mechanisms and phenotypes, when the model actually represents the disease in a clinically relevant manner.

Conditional knockout of Snrpb in the brain and neural crest lineages using the Wnt1-Cre2 driver was used to model CCMS. The heterozygous mutant embryos (Snrpb^ncc+/−) recapitulate the disease, showing craniofacial hypoplasia with decreased differentiation of craniofacial cartilage and bone, and reduced postnatal survival (Alam et al., 2022). Although most neural crest cells form in the head and migrate into the pharyngeal arches in the mutants, a subset of neural crest cells undergo apoptosis, indicating that increased neural crest cell death accounts for aspects of this disease. Snrpb^ncc+/− embryos at E9.0 also had many significantly altered splicing events compared to the wild type, with the most abundant being skipped exons and retained introns. Among these, 13 transcripts required for craniofacial development, including Rere, Dyrk2, and Pou2f1, were identified as having increased exon skipping, potentially contributing to the craniofacial defects observed in the mutant embryos (Alam et al., 2022).

The Sf3b4 ^ncc/ncc and Sf3b4 ^ncc/− mice with loss of Sf3b4 in neural crest cells is another useful model for Nager and Rodriguez syndromes (Kumar et al., 2024). This conditional knockout was able to recapitulate the craniofacial and cardiac phenotype observed in patients. Similarly, Eftud2 ^ncc−/− mouse was generated to recapitulate MFDGA in vivo. Although these mice exhibited craniofacial malformations, they did not survive until birth; however, this model was still used to understand the connection between Eftud2 and the P53 pathway (Beauchamp et al., 2021; Beauchamp et al., 2022). In particular, exon skipping and increased levels of an alternatively spliced form of Mdm2, a p53 pathway gene, were found in mouse embryos. Treatment of mutant embryos with an inhibitor of p53 (pifithrin-a) ameliorated the craniofacial abnormalities found in the untreated embryos, connecting increased p53 activity to the mechanism of MFDGA.

The mouse, however, is not always the best system for modeling craniofacial spliceosomopathies. Frequently, homozygous knockouts of splicing factors are embryonic lethal, whereas heterozygous knockouts show no craniofacial phenotype. Such is the case for the Eftud2 CRISPR/Cas9 knockout mouse, in which there is no survival post-implantation, with the heterozygotes failing to model MFDGA (Beauchamp et al., 2019). Similarly, the Sf3b4 heterozygous knockout mouse does not show any craniofacial phenotype and instead shows defects in the axial skeleton and the forebrain (Yamada et al., 2020), accompanied by mis-splicing of chromatin remodelers and dysregulation of Hox gene expression (Kumar et al., 2023). In the case of CWC27, mutant mice show the retinal degeneration phenotype of the associated disorder, but no craniofacial malformations are described in either of two mutant models—Cwc27 ^{K338fs/K338fs} and Cwc27Tm1a/K338fs (gene trapping of exon 3 and CRISPR/Cas9-mediated frameshift compound heterozygote) (Bertrand et al., 2022; Lu et al., 2023).

The best approach to study spliceosomopathies in mice appears to be the use of conditional knockouts, as demonstrated with Eftud2 and Sf3b4 using the Wnt1-Cre2 driver. However, this limits the tissues in which the defects can be examined when multiple tissues are affected. For example, conditional knockout of Eif4a3 in the radial glial cells allows for the examination of the microcephaly phenotype of patients but disregards any analysis of the craniofacial malformations observed in RCPS (Lupan et al., 2023). Therefore, considering other models in addition to the mouse may be beneficial.

Zebrafish is a popular vertebrate model system, recognized for the ease at making transgenic animals and imaging analysis due to the transparency of the embryos. Mutant embryos can be generated in a number of ways in zebrafish, with tools lending themselves to the specific attributes of a gene or disease. For example, morpholino antisense oligonucleotides can be used to target specific genes and knockdown their function. Such is the case with RNU4ATAC, in which morpholino-mediated knockdown resulted in defects in primary cilia, such as decreased number and function, thereby recapitulating the phenotype of TALS-patient fibroblasts (Khatri et al., 2023). TALEN-mediated disruption has also been used to induce mutations in zebrafish. This was done for a truncation mutation in the eftud2 gene to mimic a mutation found in a MFDGA patient. Mutants displayed a small head and small eye, identifying novel eye phenotypes possibly associated with MFDGA (Deml et al., 2015). A separate mutant construct of eftud2 known as the fn10a mutant has been generated from a mutagenesis screen, and this mutant is useful for studying the impact on neurogenesis (Lei et al., 2016). In this model, neural progenitors experience increased apoptosis and mitosis, coupled with splicing deficiencies including increased retained intron and exon skipping in genes enriched for several KEGG pathways such as “cell cycle,” “p53 signaling pathway,” and “spliceosome.” However, this mutant lacks the craniofacial phenotype of MFDGA.

Unfortunately, similar to the mouse, zebrafish models might lack certain characteristics of craniofacial spliceosomopathies observed in patients. For example, sf3b4^−/− mutant zebrafish show no craniofacial malformations but instead exhibit features of retinitis pigmentosa (Ulhaq et al., 2023; Ulhaq et al., 2024). Although retinitis pigmentosa is a spliceosomopathy (Griffin and Saint-Jeannet, 2020), it has not clinically been attributed to mutations in Sf3b4, and Nager syndrome patients do not show clinical signs of retinitis pigmentosa. Further studies will define the clinical relevance of this model.

Xenopus is an excellent model system for studying developmental disorders because of the ease at which key developmental processes can be observed and manipulated. A tool broadly used in X. laevis is morpholino antisense oligonucleotides to knockdown gene function during development. For example, a comparative study has been performed by knocking down individual splicing factors—namely, eftud2, snrpb, and txnl4a, which have been linked to MFDGA, CCMS, and BMKS, respectively—and analyzing the consequences on neural crest and craniofacial development. The main results indicate that neural crest progenitor formation is similarly affected in each knockdown through a mechanism that involves increased apoptosis and results in hypoplastic craniofacial cartilages (Park et al., 2022). A study using a morpholino against eif4a3 found that loss of this protein function disrupts derivatives of the neural plate and neural plate border, including some neural crest derivatives (although the craniofacial structure is unaffected) (Haremaki et al., 2010). Similar morpholino studies to interfere with sf3b4 or sf3b2 function have shown shared mechanisms underlying these phenotypes, characterized by impaired neural crest formation, coupled with an increase in apoptosis in the head region and reduced craniofacial cartilages, perhaps hinting at a common root cause to some, if not all craniofacial spliceosomopathies (Devotta et al., 2016; Timberlake et al., 2021).

The X. laevis allotetraploid genome and the relatively long generation time (10–12 months) render genetic analysis challenging in this organism. In recent years, the related species X. tropicalis has become more broadly used. It offers the same embryological advantages as its allotetraploid counterpart, with a shorter generation time (5–7 months) and a diploid genome. Xenopus tropicalis has been used to generate CRISPR/Cas9 knockout mutant lines, which enable a more consistent knockout than morpholinos, reducing off-target effects and allowing for large-scale genomic analysis. Recently, an Sf3b4 knockout mutant line was generated, and it was found that the homozygous null embryos showed reduced neural crest cell migration, increased apoptosis in the head region, and decreased craniofacial cartilage precursors (the heterozygous embryos were comparable to the wild type). These phenotypes were reflected in dysregulated genes as revealed by bulk RNA-sequencing, with downregulated genes categorized into GO terms such as “neural crest cell migration,” “extracellular matrix organization,” and “negative regulation of extrinsic apoptotic signaling.” These dysregulated genes were preceded in developmental time by mis-splicing events, predominately increased abnormal skipped exons, for genes categorized into GO terms such as “RNA splicing,” “regulation of embryonic development,” and “regulation of apoptotic process” (Griffin et al., 2025). Further studies will use this information to identify the gene networks and pathways affected under this craniofacial condition. These studies show that frogs provide a unique system for studying craniofacial spliceosomopathies within the context of development.

Animal models are extremely powerful tools to investigate disease mechanisms. However, there are also downsides to working with these models. Most animal models require a complete gene dosage reduction to recapitulate the disease, taking away from the ability to exactly replicate the human conditions, which are often haploinsufficient. Moreover, the reliance on conditional mutations to model a desired phenotype restricts the ability to interrogate the role of a gene in a broad range of cell and tissue types. Finally, animal models are inherently different than humans and, therefore, may introduce variables that are not directly relevant to the human diseases. Therefore, turning to in vitro modeling may help support and expand the findings of in vivo studies.

In vitro modeling of craniofacial spliceosomopathies is an emerging field, with only a small number of disorders examined so far. The use of relevant cell lines, whether primary or engineered, especially those of human origin, holds great potential for elucidating the underlying mechanisms of these diseases in terms of gene function.

For example, the use of mouse mandibular MEPA (mouse embryonic pharyngeal arch) cell lines to examine the role of RBM10 in TARP syndrome has allowed for the identification of RBM10-binding sites in the genome and the elucidation of its role in regulating alternative splicing (Rodor et al., 2016). This group also used the MEPA cells to generate an RBM10 CRISPR/Cas9 knockout cell line to characterize the phenotype at the cellular level, which highlights that loss of RBM10 leads to proliferation defects and changes in the differentiation potential of mutant cells. Human cell line HEK293 has been used to generate an EFTUD2 CRISPR/Cas9 knockout cell line, with a heterozygous loss-of-function mutation that is a null allele equivalent to MFDGA patient mutations (Wood et al., 2019). This cell line was used to identify diminished proliferation, increased sensitivity to endoplasmic reticulum (ER) stress, and mis-expression of ER stress response genes as the potential underlying mechanisms of MFDGA. Another study used primary cells—fibroblasts from TALS patients—to understand the role of RNU4ATAC in the disorder and compare function to the in vivo phenotypes (Khatri et al., 2023). The work indicates alterations in primary cilium function in these cells, which reflected phenotypes observed in vivo, thereby demonstrating the strength of this in vitro model in recapitulating some aspects of the disease.

The use of human embryonic stem cells (hESCs) and iPSCs to model diseases has grown exponentially over the recent years. These cells offer the ability to use human samples to investigate the manifestation of disorders in specific cell types or in tissues in the form of organoids. Recently, hESCs have been used to investigate the underlying mechanism of Nager syndrome. Taking advantage of a well-described protocol to derive neural crest cells from hESCs (Bajpai et al., 2010), it is possible to specifically investigate the function of SF3B4 in differentiating neural crest cells. siRNA-mediated knockdown of SF3B4 revealed a requirement for SF3B4 in neural crest cell production, survival, and differentiation (Griffin and Saint-Jeannet, 2025), showing some parallels with the corresponding animal models (Devotta et al., 2016; Kumar et al., 2024; Griffin et al., 2025). Similarly, BMKS patient iPSCs have been used to investigate the differentiation potential and behavior of neural crest cells with reduced TXNL4A expression (Wood et al., 2020). TXNL4-deficient cells exhibited defective differentiation into neural crest cells, with significant differences in neural border and neural crest marker genes, a delay in the epithelial-to-mesenchymal transition, and dampened response to WNT signaling, an important regulator of craniofacial development (Wood et al., 2020). RCPS patient-derived iPSCs have been used to generate cortical organoids to study neurogenesis with EIF4A3 haploinsufficiency (Lupan et al., 2023). Coupled with in vivo mouse work, it was determined that EIF4A3 mediates neurogenesis by controlling mitosis and cell survival; with reduction in EIF4A3, there is extensive cell death and impaired neurogenesis.

In vitro modeling of diseases allows for the use of human samples to interrogate the mechanisms of the disorders in the context of the patient mutations and/or specific cell types that are affected. In the case of craniofacial spliceosomopathies, hESCs and iPSCs can be differentiated into neural crest cells and their derivatives, allowing for examination of disease-causing mutations in the cells that are primarily affected in these patients. However, there are also some limitations to in vitro disease modeling, such as the limited number of cell types differentiating in a dish that does not fully capture the complexity of tissues in vivo; future technological advances such as 3D organoids may help alleviate some of these shortcomings. Another limitation is the difficulty of obtaining samples from patients with such rare diseases.