ARID5B mutations cause a neurodevelopmental syndrome with neuroinflammation episodes

Reviewer #1 (Public review):

Summary:

This manuscript describes a putative clinical association between ARID5B genetic variants and a novel neurodevelopmental syndrome characterized by global developmental delay, intellectual disability, and occasional neuroinflammatory episodes. While the identification of 29 individuals with overlapping phenotypes and the use of a CRISPR-Cas9 mouse model suggest a potential gene-disease link, the study suffers from significant methodological gaps in variant prioritization and a lack of robust mechanistic evidence to support its primary claims. Specifically, the "neuroinflammation" component is over-emphasized despite appearing in only a minor subset of the cohort, and the molecular pathogenesis remains insufficiently explored beyond initial protein localization assays.

Strengths:

(1) The study proposes a new clinical syndrome associated with the ARID5B gene, distinguishing it from established Coffin-Siris syndromes related to other ARID family members.

(2) The recruitment of a relatively large cohort of 29 individuals from diverse geographical and ethnic backgrounds strengthens the initial phenotypic description.

(3) The combination of human clinical data, in vitro localization assays, and an in vivo mouse model provides a multi-level framework for investigating the gene's function.

(4) The identification of variants in the exceptionally long final exon of ARID5B that escape nonsense-mediated mRNA decay (NMD) offers an interesting perspective on the molecular pathology of this gene.

Weaknesses:

(1) The description of the genomic methodology appears limited. A more detailed explanation of the filtration and selection process for variant prioritization is essential. The authors should provide a comprehensive summary of evidence (e.g., CADD scores, allele frequencies in gnomAD, and segregation analysis) to justify the selection of the reported variants, even if they do not strictly meet all ACMG/AMP criteria.

(2) The cohort includes several inherited variants and missense mutations that require more robust evidence of pathogenicity. For example, the presence of the variant in population databases (gnomAD) suggests the need for careful re-evaluation of its causality. A more rigorous assessment using diverse computational metrics, such as PhyloP scores and conservation analysis, is necessary to confirm the pathogenicity of the missense variants.

It is recommended that the authors re-evaluate the cohort to ensure that only variants with strong evidence of causality are included to maintain a clear genotype-phenotype correlation.

(3) The proposed molecular mechanism would benefit from further empirical support. The claim of NMD escape is currently supported by only a small number of cases, and a much more detailed explanation is also required for the experimental data provided.

Although the mouse model exhibits developmental abnormalities, it does not recapitulate the other systemic features reported in humans. In addition, given that "brain development" is a central theme, the manuscript lacks detailed neuroanatomical data, histopathology, or other molecular biological (e.g., RNA-seq) evidence from brain specimens to substantiate these claims at a molecular level.

(4) The emphasis on "neuroinflammation" in the title may be disproportionate to its observed frequency. Central nervous system inflammation was identified in only a small subset of the cohort (2 of 29 individuals).

Without additional experimental validation, such as immunological challenges in the Arid5b mouse model, it is premature to characterize this as a hallmark feature. Additionally, the inconsistent response to immunotherapy suggests that the autoimmune component requires further investigation.

(5) Supplementary tables require reorganization to improve clarity. The current structures make it difficult for readers to effectively analyze the data, and a more standardized format is recommended.

(6) As the manuscript proposes a novel disease entity, a more comprehensive clinical discussion is warranted. The authors should provide a more systematic description of the core clinical features and, crucially, address the genotype-phenotype correlation. Specifically, a more detailed analysis is required to determine whether the clinical severity or the presence of specific features varies according to the location of the variant or the type of mutation. Such insights are essential for clinicians to differentiate this syndrome from other ARID-related disorders.

Reviewer #2 (Public review):

Summary:

The authors compiled 29 patients with various neurodevelopmental symptoms due to the ARID5B mutations. Although not directly, the mouse model demonstrated that the heterozygous mutant mouse showed mild behavioral problems. It would be interesting to see if the mice carry craniofacial features.

Strengths:

The HEK293T model showed that the mutant protein mis-localized, but did not show whether the mutation caused any changes in epigenetic status. Nevertheless, this paper delivers clear support for genotype-phenotype correlation.

Weaknesses:

(1) The paper would be improved by providing pedigrees of some of the patients with inherited variants.

(2) Figure 3d could provide more species for an accurate conservation assessment.

Reviewer #3 (Public review):

Summary:

In the present study, through international gene-matching efforts, the authors present 29 individuals with rare, heterozygous ARID5B variants and find that these individuals have a newly recognizable neurodevelopmental syndrome. A recurring clinical syndrome of developmental delay/intellectual disability, behavioral difficulties, renal malformation, and recurrent infections is described. 19 of these variants were confirmed to be de novo, and only one was inherited from an unaffected parent. 24/29 of these variants introduce premature termination codons in the final exon and are predicted to escape nonsense-mediated decay. The ARID5B p.Q522Ter variant was studied in a mouse heterozygous knock-in model, found to be associated with behavioral abnormalities. The well-described genetic and phenotypic data for this cohort provide convincing clinical evidence for a novel neurodevelopmental syndrome. The functional evidence provided is preliminary, and further studies are needed to understand disease mechanisms.

Strengths:

(1) The authors give a good description of a novel clinical syndrome manifesting as developmental delay/intellectual disability, facial dysmorphism, and behavioral challenges.

(2) The authors create a mouse model harboring an Arid5b(Q522*/+) variant and identify subtle behavioral changes.

(3) Attempts are made to functionally characterize a subset of ARID5B variants in human cell lines.

Weaknesses:

(1) The title - "ARID5B mutations cause a neurodevelopmental syndrome with neuroinflammation episodes" - should be revised. 2/29 individuals (7%) had CNS inflammation; this does not appear to be a core feature of the disease and should not be highlighted as such. If this is going to be a feature that is highlighted, then more details are needed. MRI images of cerebellitis and/or ADEM would be helpful, as well as lumbar puncture results and supplemental information detailing the treatment course.

(2) The abstract states that "Remarkably, 19 of 29 variants (66%) cluster within the first quarter of exon 10, are de novo, and escape nonsense-mediated mRNA decay (NMD), which we confirmed for two variants affecting seven individuals." The authors state in the Results that they "indeed found no signs of NMD". In Figure 3f, when assessing for transcript amount, there appears to be a great deal of variability. Three ARID5B variant lines are tested. Transcript amounts in two lines appear to be near control levels, but one LCL ARID5B Ile497AsnfsTer31 line appears to demonstrate significantly lower levels of transcript. The control lines also show a great deal of variability. No explanation is given for this large difference between LCL ARID5B Ile497AsnfsTer31 lines and for the variability in control lines, making these data uninterpretable. A major theme of the paper is that early truncating variants in exon 10 escape NMD and lead to the described phenotypes, so this is an important point that needs to be resolved, either by testing more patient-derived lines or knocking in these variants into cell lines.

(3) The Arid5b(Q522*/+) mice are not sufficiently molecularly characterized. Does the variant transcript escape NMD? What happens at the protein level? Is there mislocalization of the protein?

(4) For the HEK293T cell experiments, variants are overexpressed and compared to a control. These experiments appear to leave endogenous ARID5B intact. What might the authors expect to see if these variants were knocked in?

(5) The functional consequences of the missense variants are not tested. The authors suggest that missense variants may be more associated with macrocephaly and possibly ASD. Are these missense variants causing loss-of-function or gain-of-function? Is there preserved protein function?

(6) There are a number of functional assays performed, but it remains unclear if the tested variants are operating through a loss- or gain-of-function. Are truncating variants early in exon 10 leading to a partial loss-of-function? Or do they prevent the functioning of the other allele through a dominant negative mechanism? These possibilities are not directly tested.