Epigenetic diseases that are caused by genetic variants have emerged as a major category of neurodevelopmental disorders (NDD). The AT-Rich Interaction Domain 5B (ARID5B [OMIM: 608538]), known as modulator recognition factor 2 (MRF2) or DESRT, belongs to the ARID transcription factor family which have key roles in gene regulation via chromatin remodeling and epigenetic modification


1


. Variants in several members of the ARID family have been associated with neurodevelopmental disorders, in particular

ARID1A

(Coffin-Siris Syndrome 2 (CSS2, [OMIM #614607]),

ARID1B

(Coffin-Siris Syndrome 1 (CSS1, [OMIM #135900])


<sup>2

,

3</sup>


and

ARID2

, another member of the BRG1-associated complex (BAF) chromatin remodeling complex, leading to a Coffin-Siris-like syndrome


⁴


. ARID family members that are not members of the BAF complex have been less frequently associated with NDD. How-ever, via meta-analyses of a genomic study on intellectual disability (ID)


⁵


, we could associate a loss of function variant (LoF) in

ARID5B

with NDD. Notably, this variant would escape nonsense-mediated mRNA decay (NMD) as a consequence of the premature termination co-don (PTCs) being in the proximal part of the exceptionally long last exon, exon 10 of

AR-ID5B

(2.16 kb, 98th percentile for exon length based on NCBI human exon sizes)


<sup>6

,

7</sup>


.

Through gene matching, we initially identified three additional individuals with

de novo ARID5B

variants presenting with intellectual disability, speech and global developmental delay, learning difficulties and two with kidney defects. These individuals did not resemble the set of phenotypes typical of Coffin-Siris(-like) syndromes and lacked any disease-causing variants in known neurodevelopmental disorders. Moreover, all three truncating variants clustered in the first quarter of the long exon 10, similar to the reported variant, which prompted us to systematically characterize the clinical manifestations associated with

ARID5B

disruption.

Here, we present a cohort of 29 probands with developmental delay and rare ARID5B variants identified via the Matchmaker Exchange platform


⁸


. The individuals have divergent geographical and ethnic backgrounds and females and males are nearly equally distributed. Remarkably, the majority of individuals have truncating variants with premature codons in the long exon 10 that affect both TV1 and TV2 transcript variants while escaping NMD. The others four variants are missense variants in conserved missense intolerant domains in the N-terminal half of the protein.

Materials And Methods

Cohort selection criteria

Inclusion criteria were individuals with a predicted deleterious (premature termination codon or missense) variant in

ARID5B

and a neurodevelopmental disorder (NDD), in the absence of known NDD causing variants. The criteria for a NDD were either if patients presented with global developmental delay (all but one), or when speech was delayed, or when intellectual disability was diagnosed (ID). All the cases were isolated, except for three sisters, who inherited the variant from an affected mother. All variants were identified through exome or genome sequencing.

Data and sample collections

Informed consent for genetic studies as well as the blood sampling and publication of photographs included here was obtained from parents or legal guardians. Written, signed forms were accompanied for consent to use the donated cells and images of the participating individuals or their parents. Lymphoblastoid cell lines (LCLs) were provided by Care4Rare Canada. Peripheral blood mononuclear cells (PBMCs) controls were generated following blood sampling and from cohort individuals provided by M.D. and R.J. and subsequently processed at the Centre hospitalier universitaire Sainte-Justine (CHUSJ) Institutional Mother-Child Biobank, following standard protocols.

Mice

C57BL/6NJ mice were maintained in an established facility at the Centre de recherche Azrieli du CHUSJ (CRA-CHUSJ). Mouse husbandry and colony maintenance were performed according to the animal protocol approved (2022-3860) by the “Comité Institutionnel des Bonnes Pratiques Animales en Recherche” (CIBPAR). This committee is following the guidelines of the Canadian Council on Animal Care (CCAC).

Generation of Arid5b


^emQ522*


mice

The human variant NM_032199.3 c.1564C>T, p.Gln522Ter, was introduced into the mouse genome using CRISPR/Cas9. The mice were generated by the McGill Integrated Core for Animal Modeling (MICAM; McGill University, Montreal, Quebec, Canada). Briefly, the sgRNA (Synthego), Cas9 protein (IDT, catalog 1081058) and ssODN (ultramer, IDT) were microinjected into the pronucleus of C57BL/6N mouse zygotes with respective concentrations of 50:50:30 ng/μL. Embryos were subsequently implanted in CD-1 pseudopregnant surrogate mothers according to standard procedures approved by the McGill University Animal Care Committee (UACC). Oligonucleotides used were mArid5b-gRNA: 5’-
AAGGCCAATGAAACTGACCA
-3’ and ssODN:5’-
TGGGAGCTGAATCTTTTTCAGGAGGCAAGGGAGGGCTTGGAAGCAGAGGGGCAA
GGCCCTTATCTCCCATCTCCTCGGCCTCTTTCTCGCTGTTGGAA

gg

TT

aa

TCAGTTT
CATTGGCCTTTTCTGGGTCTGCTCTGGACAC
-3’. After weaning, the mice were transferred to the CRA-CHUSJ. Mice were backcrossed for at least 3 generations to C57BL/6NJ (Jackson). The presence of the variant was confirmed by Sanger sequencing. Mouse husbandry and experiments at CRA-CHUSJ were done according to the approved animal user 732-NAGANO protocol no. 2021-3228 by the coordinator of the CIBPAR, in accordance with the CCAC guidelines.

Staining quantification

ImageJ software


⁹


was used to quantify pictures of fluorescence semi-automated and blinded. For each data point in

Figure 5B

, 2 wells of a 12-wells plate were used per condition and for each well, 2 images were generated and quantified. For the quantification in

Figure 5B

, channels from the overlay ImageJ.tif files for DAPI and FLAG were split. The option Process>Binary>Make binary was used and ‘Analyze Particles’ to determine the region of interest (ROI). For the settings, a range of 20 to 1000000 pixels for FLAG and from 10 to 1000000 for DAPI was chosen and holes were included. This way, we verified that almost solely separated single nuclei were selected using the DAPI channel and FLAG staining images generated ROIs that corresponded to single cell expression, regardless of whether the staining was cytosolic and/or nuclear, this way selecting transfected cells. Next, the nuclear ROI was overlayed on the 16-bit FLAG-channel layer to quantify total FLAG staining per cell. Next, the DAPI ROI overlay was again used to fill up the nuclear ROI within the FLAG-16 bit image, masking the nuclear FLAG-staining but leaving the cytosolic expression. Then, we used the FLAG ROI again to quantify single cellular cytosolic staining. Nuclear staining per individual cell was determined as the difference between total FLAG-staining and cytosolic staining. Finally, each image output nuclear/cytosolic ratios for each cell were calculated per cell, generating several hundreds of measurements per image, then averaged per well and used for statistics and graphical presentation. This was repeated for three separate experiments and the averages of each experiment are plotted and used for statistical analyses. For the quantification in

Figure 5G

, one image per (n) per condition was taken and measurements were based on the size of the FLAG area per cell (imageJ, ROI per cell), see Rscript. Each data point is the mean of ROI surfaces of all individual cells in one coverslip/12-wells well. The wells were from three separate plates.

See

supplemental information

for additional, more standardized experimental procedures and statistics

.

Results

De novo

truncating

ARID5B

variants cause neurodevelopmental phenotypes

Assessing the potential deleterious effects of variants in the ARID family, other than those in the previously established critical neurodevelopmental genes

ARID1A

and

ARID1B

, we initially were compelled by a loss of function (LoF) variants in

ARID5B

in a genomic study on individuals with intellectual disabilities


⁵


. We next collected information on three additional individuals with developmental delay (DD) and

ARID5B

variants (not selecting for a specific type of variant) with exome or whole genome sequencing. Remarkably, all four variants caused premature termination codons (PTCs) at the beginning of the long final exon 10, leading to the truncation of at least half of the protein while predicted to escape nonsense-mediated mRNA decay. This prompted our hypothesis that truncating variants at the beginning of exon 10 can cause neurodevelopmental perturbations and motivated us to collect a larger cohort of

de novo ARID5B

variants, for which we also aimed to collect a more complete clinical profile.

With the help of Matchmaker Exchange platform tools


<sup>8

,

10</sup>


we have assembled a cohort of in total 29 probands with neurodevelopmental phenotypes and

ARID5B

variants of which 24 (83%) are loss of function variants that have PTCs in exon 10 and one that affect the exon 9/10 splice site. Four others are missense variants in conserved protein domains in the N-terminal half of the protein. (

Fig. 3B,C

). Nineteen (68%) variants lead to gained stops or frame shifts in the first quarter of the long last exon 10, all of those that were assessed (14 individuals) were

de novo

variants. However, five out of six exon 10 truncating variants that were more dispersed over the other three quarters of exon 10 were inherited from three symptomatic mothers (including three sisters) and one non-symptomatic mother. Inheritance of a sixth C-terminal truncating variant could not be assessed.

A detailed summary of overlapping dysmorphisms and images can be found in

Figure 1

. In

Figure 2

a detailed summary of frequencies of other overlapping phenotypes can be found. Both the number of individuals and the percentages are noted, the latter are calculated by excluding individuals of whom a particular phenotype was not assessed at the time of writing. Detailed per patient information including phenotypes that occur only once can be found in

Supplementary table 1

.

Of the 29 probands, 14 females and 15 males with the age of their last follow up ranging from 0 to 29 years, all presented with a neurodevelopmental disorder. All individuals had global developmental delay (GDD), with the exception of one individual who was not assessed but suspected to have mild ID. Ten individuals were officially diagnosed with intellectual disability (ID [OMIM: 156200]), mostly mild (seven individuals), eight others likely have ID, but were not formally diagnosed at the time of writing, and all others were suspected to have cognitive challenges. Behavioral issues were frequent (18 individuals), including autism spectrum disorder (ASD [OMIM: 209850]) (8 individuals) and other behavioral issues (14 individuals), particularly attention-deficit/hyperactivity disorder (ADHD [OMIM: 143465]). Speech was affected in 21 individuals, including one individual with regressive language externalization leading to speech apraxia.

Frequent systemic manifestations that characterize the disease are kidney abnormalities (11 individuals) including hydronephrosis (five individuals), megaureter (two individuals), pyelectasis and pyelic hypotony. Hypotonia (10 individuals) and feeding difficulties (nine individuals) were common, with three individuals presenting with oropharyngeal dysphagia. Visual impairments (seven individuals), including myopia (OMIM: 160700) (6 individuals) and in addition three individuals with strabismus (OMIM: 185100). Ear/auditory issues (seven individuals) included sensorineural hearing loss (2 individuals) and tympanostomy tube placement (two individuals).

Immune dysfunction was noted in nine individuals, mostly with recurrent infections (seven individuals), two individuals presented with rare forms of CNS inflammation and another individual was hospitalized for a prolonged fever with unknown cause. Individual 27 (8-year-old girl) developed autoimmune cerebellitis and was unresponsive to therapy, while individual 23 (7-year-old boy) had recurrent acute disseminated encephalomyelitis (ADEM), but responded to therapy. Both were treated with corticosteroids and intravenous immunoglobulin (IVIG) and experienced seizures related to neuroinflammation. Two infants presented with persistent pulmonary hypertension, which was the predominant cause of death in individual 19 at six months of age. Finally, out of the four individuals that had missense variants, two presented with macrocephaly and another one with (mild) microcephaly with wide ventricles and corpus callosum atrophy that appeared after the age of 10 months. Macrocephaly was not reported for truncating variants, while individual 15, a two-year-old girl with a truncating variant, presented with a head circumference <0.4th centile. Finally, intrauterine growth retardation was reported 4 times, while post-natal growth was normal, with two individuals having a tall stature, and most individuals with truncations at the beginning of exon 10 being on the tall side.

A very common craniofacial feature was the shape of the nose, generally shorter and broader, a bit bulbous, often with a lower nasal bridge (13 individuals) (

Figure 1A,B

). 10 individuals had an uncommon head shape, with a more prominent forehead and narrow face, frequently with micro/retrognathia (8 individuals). Five individuals presented with hypertelorism, ptosis and, nine individuals presented with thin upper lip, long philtrum, or narrow mouth. Skeletal abnormalities included toe/foot anomalies (7 individuals), radially deviated wrists (2 individuals), long and slender fingers or clino-comptodactyly (6 individuals) and abnormal palmar creases (6 individuals).

Premature termination codons cluster in the first quarter of exon 10 and escape NMD

Nineteen out of 29 variants (66%) cluster in exon 9 (only frameshift) and early exon 10 (frameshift and nonsense), with all PTCs being in the first quarter of exon 10 and one variant that affects the exon 9 / exon 10 splice donor site (

Fig. 3B

). Since PTCs in such exceptionally long exon 10 (2.16 kb, >95th percentile for length)


⁶


are known to escape NMD, we quantified and sequenced the exon 9 – exon 10 border at the mRNA level to confirm the absence or strongly reduction of NMD in two lymphoblast cell lines (LCLs) and primary peripheral blood mononuclear (PBMCs) of two patients generated from blood collected at different ages. The variant affecting the donor of the LCLs was five times recurring in our cohort (p.Ile497AsnfsTer31). We indeed found no signs of NMD (

Fig. 3E, F

). Unlike our cohort, gnomAD LoF variants are dispersed, with only 3/23 (13%) clustering in the first quarter of exon 10, compared to 19/24 (79%) in our cohort (

Fig. 3B

)


¹¹


. Moreover, all tested probands (5 out of 6, of which three sisters) were inherited, suggesting that the approximate location of the PTC could be relevant for the phenotypic outcomes. Less truncated, longer ARID5B protein could lead to milder phenotypes if they contain additional translated C-terminal domains that are still functional. One particular protein domain between amino acid 610-650, that is intolerant to missense variants, is lost in all truncations that cluster at the beginning of exon 10 (

Fig 3B,C

) but still encoded in all inherited truncations that affect the latter three quarters of exon 10. Moreover, the regional clustering of pathogenic LoF variants mirrors that seen in other epigenetic regulators (e.g.,

KAT6B

[OMIM: 612990],

ASXL1

[OMIM: 612990],

ASXL3

[OMIM: 615115]).

The four missense variants in our cohort resided in highly conserved, missense-intolerant regions and were not found in gnomAD V4; p.Ile54Met (BAH domain, but unknown target), pArg186Trp (domain of unknown function), p.Ser266Asn (domain of unknown function) and p.Arg335Ser (BRIGHT/ARID, DNA-binding domain). Especially the latter is in a strongly missense intolerant domain, and loss of a positively charged arginine at the end of the first alpha-helix in the the ARID5B DNA-binding domainis likely to affect binding to the negatively charged DNA


¹²


. All were conserved from mouse to human, and all but p.Ile54Met from mouse to zebrafish (

Fig. 3C,D

). Notably, in gnomAD V4


¹¹


, the p.Ser266Gly substitution can be found 12 times.

Eight individuals presented with ASD, with only four of those being truncating at the beginning of exon 10, while three out of four missense variants, among which those residing in the BAH and BRIGH domain, were associated with ASD. We also noted two ASD-associated missense variants (p.Ala702Pro, p.Thr906Ala) in genomic studies


<sup>13

,

14</sup>


. Eventhough p.Ala702Pro was absent in gnomAD, p.Thr906Ala was identified four times in gnomAD V4. Another ASD-associated variant (e.g., p.lIe213HisfsTer16) affected only transcript variant 1 (TV1) (

Fig. 3B

).

Arid5b

^(Q522*/+)


mice exhibit developmental and behavior phenotypes

To further establish deleterious developmental consequences of truncating

ARID5B

variants, we generated knock-in mice that are carriers of the truncating variant of individual 20, a gained stop at the beginning of exon 10. Following repeated backcrossing and Sanger sequencing validation of the mutation and the ends of the donor DNA arms (

Fig. 4A

), we first tested the viability of mice heterozygous and homozygous for the variant. We did not observe any effect on the one-year survival rate of heterozygous mice, but homozygous mice generally died shortly after birth (

Fig. 4B

). Of note, with manual supplementation of food, some mice were able to survive slightly longer. Since homozygous mice remained very small and vulnerable, we decided not to proceed with their characterization for ethical reasons.

Even though heterozygous mice showed no signs of reduced one-year-viability, their lower weight throughout development suggested a developmental defect that seems to recover into adulthood (

Fig. 4C

). In addition, heterozygous mice showed different behavior in the open field test, being more frequently in the center than wild type mice (

Fig. 4D, E

). Similar patterns of behavior are seen in other mouse models of ID or ASD, like Fragile X syndrome (

Fmr1

disruption), Bosch-Boonstra-Schaaf Optic Atrophy Syndrome (

Nr2f1

disruption),

AUTS2

-related syndrome (

Auts2

disruption)


<sup>15

–

17</sup>


. However, we did not observe sociability or social novelty defects, as heterozygous mice showed a normal preference for a mouse over an object and for a novel mouse over a familiar mouse in the 3-chamber sociability test(

Fig. 4F

). Also the overall time that

Arid5b

^(Q522*/+)


mice spent on social exploration did not significantly change (

Fig. 4G

).

Truncated ARID5B delocalizes into the cytoplasm

To investigate the protein consequences of truncating variants in exon 10, we constructed open reading frames (ORFs) encoding variously tagged ARID5B variants (see Methods). As expected, overexpression of ARID5B in HEK293T cells resulted in near exclusive nuclear localization under standard growth medium conditions, regardless of the tag added (

Fig. 5A

). However, introducing the nonsense variant of individual 20 (c.1489dupA, p.Gln522Ter) resulted in near complete delocalization of the truncated protein in the cytoplasm (

Fig. 5A

). Surface plots overlaying FLAG and DAPI staining further showed the strong extra-nuclear localization together with the nuclear absence of the p.Gln522Ter variant in more detail (

Fig. S2A

).

C-terminal truncating variant shows only a very minor delocalization phenotype

We set up a highly-sensitive, semi-automated blinded quantification of the cellular localization and assessed p.Gln522Ter and the most-C-terminal variant, p.Ala1146CysfsTer15. Whereas p.Gln522Ter showed, as expected, a strong cytosolic delocalization, p.Ala1146CysfsTer15 only showed a very minor but reproducible, cytosolic localization defect (

Fig. 5B

). However, we want to note that the typical localization patterns of p.Ala1146CysfsTer15 is similar to those of wild-type ARID5B, regardless of total protein expressed per cell (

Fig. S2B

), a small portion of cells showed cytosolic localization more similar to p.Gln522Ter (

Fig. S2C

). This could explain the small effect on overall cellular localization of p.Ala1146CysfsTer15 observed in

Figure 5B

.

The long exon 10 harbors 3 ATG sites that have predicted functional Kozak-sequences that have the potential to produce alternative isoforms (

Fig. S3A

). In mouse embryonic brain lysates, proteins of the expected sizes of these alternative exons react with anti-ARID5B antibodies (

Fig. S3B

). In addition, we have found evidence for alternative isoform expression from C-terminally tagged, overexpressed ARID5B (

Fig. S3C

). To investigate if such isoforms would localize in the nucleus, which would suggest the use of a C-terminal nuclear localization signal (NLS), we overexpressed this predicted alternative isoform III and assessed its localization. Indeed, this C-terminal part of ARID5B alone, encoded solely by exon 10, also resided in the nucleus (

Fig. 5C

). A detailed description of potential alternative isoforms can be found in the supplemental information and figure

Fig. S3

.

Various tools predict a complex nuclear export/import domain in exon 10

To predict protein sequences that could play a role in ARID5B cellular localization, we combined the output of several motif-predicting tools (

Fig. 5D

). NLStradamus

homo Sapiens

predicted two significant nuclear localization (NLS) sites in the N-terminal half of the protein (

Fig. 5D

)


¹⁸


. However, a broader almost significant C-terminal domain was indicated by NLS-Mapper, Prosite and MotifScan, with two NLSs flanking a nuclear export signal (NES) in the region covering amino acids 999-1030 of ARID5B isoform-I, potentially forming a complex NLS/NES/NLS (NNN) domain (

Fig. 5D

)


<sup>19

–

21</sup>


. Notably, one of the missense variants in our cohort (p.Ser266Asn) resides on the flank of the predicted N-terminal NLS (

Fig. 5D

).

Considering the truncating variants, most truncations lead to the complete loss of the C-terminal predicted NNN, including p.Tyr968Ter, and thus we hypothesized that cellular localization for variants that lost the NNN would be similarly affected (

Fig. 5D

).

Cohort and designed ARID5B variants suggest a dominant C-terminal nuclear localization domain

We generated nine more ARID5B variants tagged with a FLAG peptide to investigate conse-quences on cellular localization (

Fig. 3E

). First, we confirmed that all overexpressed truncat-ed variants result in proteins of the expected size (

Fig. S2D

). In addition to p.Gln522Ter and the more C-terminal p.A1146CfsTer15, we also included shorter and intermediate truncated variants from our cohort (c.1423C>T, p.Gln475Ter and c.2904C>A, p.Tyr968Ter) and the NLS-located missense (p.Ser266Asn). Further, we added several designed variants aiming to explore the mechanism of cytosolic localization in more detail, including one with a small de-letion (p.del1018-1026) in the middle of the NNN domain and additional missense in the primary NLS signal (p.Ser264/Ser266Asn) that affects a potential regulatory serine in the N-terminal NLS domain near the variant of individual 3. We also added one C-terminal truncat-ing variant from gnomAD V3 (p.Glu634AlafsTer29) and one missense variant derived from the genomic ASD study, p.Ala702Pro


¹⁴


. As expected, truncating variants from the cohort (p.Gln474Ter, p.Tyr968Ter) and gnomAD truncating (p.Gly634AlafsTer29) variants that completely lose the NNN domain located in the cytosol (

Fig. 5F

). We further quantified this by measuring the (increased) average surface expression per cell (

Fig. 5G

). Of note, we could not find any changes in expression patterns caused by stress or altered growth conditions fol-lowing temporarily depleting cells of serum (

Fig. 5E

).

The p.Ser266Asn cohort variant located in the N-terminal NLS did not lead to more FLAG-tagged protein outside the nucleus. Additional mutation of p.Ser264Leu (p.Ser264Leu/Ser266Asn) did not change this finding (

Fig. 5F,G

). Strikingly, a small deletion in the middle of the NNN (p.del1018-1026) had a surprisingly strong effect, completely delocalizing ARID5B in the cytosol (

Fig. 5F,G

). Other designed variants with altered charges of the N-terminal NLS (p.Lys255Asp/Lys256Asp) and one variant with two serines mutated in the first NLS of the NNN domain (p.Ser1002/1015Asp) did not change the cellular localization (

Fig. 5F,G

). Neither did the ASD-associated p.Ala702Pro variant affect the nuclear localization of ARID5B, all having a punctate nuclear signal that is typical for wild type ARID5B (

Fig. S2E

). In summary, our data reveals a conserved C-terminal ARID5B protein domain that can regulate ARID5B cellular localization, explaining cellular delocalization of truncated ARID5B variants.

Discussion

We have gathered evidence that truncating

ARID5B

variants, especially those that cluster within exon 9 and at the beginning of the exceptionally long last exon 10 of

ARID5B

, escape NMD, cause a neurodevelopmental disorder with GDD, ID and speech delay. Additional features included kidney defects and recurrent infections. Two individuals had episodes of CNS inflammation and two infants had persistent pulmonary hypertension, being the main cause of death of a 6 months old boy. Facial dysmorphic features most frequently include a short, bulbous nose, noted for half of the individuals, often accompanied by a shorter/lower nasal bridge. Several individuals presented with micrognathia, ptosis and some with a prominent forehead.

While none of the variants that cluster in the first quarter of exon 10 were inherited (14/19 were tested), three out of four of the truncating variants outside the cluster were inherited and the fourth one could not be addressed, which suggest that these variants could have unique deleterious effects for some phenotypical aspects, possibly with milder ID. While one mother was not diagnosed, two others were affected.

Truncations that lead to the loss of fewer domains may lead to reduced loss of function compared to variants that loos the complete exon 10 coding coding RNA, particularly if the additional domain(s) are still functional. Since LoF variants that lead to complete NMD are so rare, as discussed above, we opt that the lack of losing a conserved domain of unknown function, spanning amino acid 610 and 650 of ARID5B isoform one, can play a role. This domain may still be intact in all inherited truncating variants, while being lost in variants that cluster at the beginning of exon 10.

Another possible mutation type specific phenotype could be macrocephaly, with which two individuals with missense variants presented but none of the individuals with truncating variants. Also, four out of eight individuals with ASD had variants that did not cluster in exon 10, while three out of four individuals with a missense variant had ASD. Several missense and one LoF of TV1 were identified in individuals with ASD. We did not find clear other patterns in genotype-phenotype relationships.

None of the missense variants were identified in gnomAD V4, and the affected residues are conserved between mouse and human (p.Ile54Met, individual 1), or between zebrafish, mouse and human (p.Arg186Trp, p.Ser266Asn, p.Arg335Ser, indivduals 2-4). All were present in missense intolerant regions, and three were confirmed to be

de novo

, suggesting (neuro)developmental importance. However, p.Ser266Asn can be found substituted into a glycine (12X) in gnomAD V4 and had unique phenotypes (no ASD, mild microcephaly) compared to the other missense variants. The other three individuals with missense variants and ASD, have variants residing in known critical domains like the BAH and ARID domain, and two of them presented with macrocephaly (BAH and ARID domain variants).

ARID5B

is predicted to be haploinsufficient considering the gnomAD V4 database, having a loss-of-function (LoF) observed/expected upper bound fraction (LOEUF) of 0.21, with the current threshold for V4 being 0.6


¹¹


. However,

ARID5B

variants that are predicted to lead to the complete loss of function (LoF)

ARID5B

via NMD of both

ARID5B

transcript variants (TV1 and TV2) are much more sparse than LoF in general. To our knowledge, out of >800.000 individuals in gnomAD V4, and theoretically dozens more that could have been identified through gene matching, we could only find two individuals (both in gnomAD V4) that carried variants that are predicted to lead to NMD of both

ARID5B

TV1 and TV2. In addition, variants in the exceptionally long final exon, exon 10, generally escape NMD


⁷


.

Therefore, the translational consequences can be particularly divergent for

ARID5B

, with LoF variants in a large part of the coding regions escape mRNA decay, or otherwise, only affecting TV1, not TV2, leading in both cases to only a partial loss of endogenous ARID5B protein functions and/or abnormal protein variants.

Both complete (TV1+TV2) LoF variants were identified in gnomAD, while in our cohort, and in many large genomic ASD and ID studies, no additional similar variants were identified. This extremely low frequency could emphasize a strong deleterious effect compared to other

ARID5B

variants, perhaps generally too strong to be compatible with survival. The presence of variants in gnomAD is sometimes interpreted as meaning that they are unlikely to cause neurodevelopmental disorders, but at very low frequencies, one should be cautious interpreting gnomAD in such a way


²²


–it should be kept in mind that these individuals for instance may have a mosaic variants, or are affected by these variants. We conclude that only complete genotype and phenotype characterization of heterozygous individuals with LoF variants can determine if such ultra-rare variants indeed lead to limited or variably penetrant phenotypes. For now, we expect that complete

ARID5B

LoF (affecting both transcript variants) will generally be as deleterious, or more deleterious as the truncating variants described here.

A few other potentially deleterious

ARID5B

variants have been previously described. One individual with a nonsense variant affecting only TV1 (p.Gln84Ter) presented with a neurodevelopmental disorder, with ID, in combination with seizures, pituitary defects, a severe short stature


²³


, which are mostly phenotypes we have not observed with other

ARID5B

variants. Several other variants, both LoF affecting only TV1, and missense variants in the C-terminal half of ARID5B, have been associated with autism spectrum disorder (ASD)


<sup>13

,

13

,

14</sup>


and several more can be found in gnomAD V4.

Several immune and inflammatory roles have been ascribed to

ARID5B

. For instance, in adaptive natural killer (NK) cells,

ARID5B

mRNA levels correlate with the expression of electron transport chain genes, oxidative metabolism and Interferon gamma levels


²⁴


.ARID5B’s role as an immune mediator has been supported in aged monocytes, in which increased

ARID5B

expression correlates with atherosclerosis. In these aged monocytes,

ARID5B

is upregulated by immune activation and acts upstream as an activator of lipid metabolism and inflammatory genes. Moreover, the knockdown of

ARID5B

also reduced their migration, while increased expression correlated with increased mobility


<sup>25

,

26</sup>


.

ARID5B

transcription itself is under the influence of

ARIEL

(

ARID5B

-inducing enhancer-associated long noncoding RNA/

XLOC_005968)

which is also up-regulated in TAL1-positive types of T-ALL and acts as an enhancer RNA by facilitating

ARID5B

enhancer-transcription complexes


²⁷


. A SNP in intron 4 of

ARID5B

, wherein two alternative exons reside


²⁸


, has been linked to autoimmune diseases, specifically rheumatoid arthritis (RA [OMIM: 180300]) and Graves disease (GD [OMIM: 275000])


²⁹


. In our cohort, two individuals had severe neuroinflammatory episodes (individuals 23 and 26) and a third individual (20) showed regressive loss of speech externalization from the age of two and was hospitalized for several weeks due to a prolonged fever with unknown cause around the age of 10. Given the rarity of cerebellitis and encephalomyelitis in pediatrics, the often precarious clinical situation that may arise, and given the numerous associations between

ARID5B

and immune system function, we think that it is critical to emphasize that truncating

ARID5B

variants are likely causally related to the recurrent CNS inflammatory episodes and other immune perturbations experienced by this subset of patients.

Furthermore, we discovered that the long exon 10 ORF has an independent nuclear localization domain that is lost for most variants following truncation, except the two most C-terminal variants. These former variants localize in the cytosol, and not in the nucleus of HEK293T cells, in which ARID5B is normally found with a typical punctate nuclear localization (

Fig 5

,

Fig. S2

). We observed this effect particularly for variants that have translation termination codons upstream of a predicted, complex nuclear localization sequence (NNN). The NNN is very likely involved in ARID5B nuclear localization, in particular because we could drastically diminish ARID5B nuclear localization by a small, designed deletion in the center of the broad NNN domain. Moreover, the predicted isoform-III, which lacks two predicted NLSs that are present in the former half of the main ARID5B isoform, localized in the nucleus, while the loss of the last 42 amino acids (p.A1146CfsTer15, individual 29) of the C-terminus did not substantially affect cellular localization (

Fig. 5B,F,G

). Still, it remains uncertain if the removal of the center of the NNN is directly involved, or that it affects protein localization indirectly by changing the (local) protein conformation. Furthermore, since serine phosphorylation often plays a role in nuclear localization, for example of IKAROS, an ARID5B coregulator


<sup>30

,

31</sup>


, we substituted two serines in the NNN-domain but that did not affect the cellular localization of ARID5B.

Escaping

ARID5B

NMD may affect its function in at least three ways as compared to complete LoF. Firstly, the loss of domains encoded by exon 10 may cause only a partial LoF, with the truncated protein still executing some of its functions. Considering the genotype-phenotype relationship, we would like to emphasize that truncating variants that cluster in the first quarter of exon 10 lose all three distinguishable conserved protein domains (or regions) encoded by exon 10 (

Fig. 3

), while the most C-terminal variants may not lose the function of at least two of these domains. Secondly, the truncated protein may have gained functions harmful to neurodevelopment. Thirdly, transcripts that bypass NMD may still express the alternative isoform(s) encoded by exon 10, and their functions may therefore not be lost, reducing the deleterious effect of LoF variants. Investigating the protein localization of ARID5B, isoform expression and truncated protein expression in other cell types and our mice, would surely help address these questions, but are currently challenged by a lack of antibodies that effectively bind to the N-terminal half of ARID5B.

Three separate mouse models with deletions of exons encoding portions of the BRIGHT domain showed strain-dependent sub-lethality frequencies, growth retardation and were lean


<sup>32

–

34</sup>


.

Arid5b

mutant mice only deviate postnatally in weight, as adults when they become smaller than wild-type mice


³³


. Both homo- and heterozygous

Arid5b

mutant mice are resistant to high-fat diet-induced weight gains. Such resistance to diet-driven weight gains has also been observed in mice lacking ARID5B solely in fat cells by Fabp4-Cre induced deletion of exon 6


³⁴


, which seems to be the consequence of triglyceride and lipid metabolism being downstream of ARID5B


<sup>34

,

35</sup>


. Even though, in humans, a SNP in an ARID5B motif has been associated with obesity, and mice lacking ARID5B are lean (or at least are protected against diet-induced weight gains)


<sup>34

–

36</sup>


, we did not observe clear trends in weight or growth defects in our cohort, with the individuals falling within a broad spectrum of weight and height. However, 4 individuals presented with IUGR, while later in life, individuals may catch up or even have a taller than average stature. In our mice, their reduced birth weight is at least partially compensated for into adolescence as well. Heterozygous mice seem to catch up, while homozygous mice are too vulnerable to make it through postnatal phases without support.

This discrepancy can be due to several factors. Firstly, the human SNP may affect the binding of other transcription factors, not only ARID5B function. Furthermore, mice that lack ARID5B completely likely have unique phenotypes that could differ from variants that are truncating and escape NMD and/or have gain-of-functions. Therefore, comparing between truncating variants, NMD-causing ‘gene-trap’ mice, or overexpression models should be done cautiously.

We also found subtle changes in the open field test, with, perhaps unexpectedly, the mice spending more time in the center of the arena in the open field test, a phenotype also observed for other ASD/ID NDD mouse models, for example Fragile X syndrome (

Fmr1

disruption), Bosch-Boonstra-Schaaf Optic Atrophy Syndrome (

Nr2f1

disruption),

AUTS2

-related syndrome (

Auts2

disruption)


<sup>10

,

16

,

17</sup>


, and in other mouse models that have established cognitive limitations


³⁷


. An overall lack of awareness, eventual in combination with a lower overall drive to explore could prompt spending more time in the center of an open field, likely even in the absence of an anxiolytic phenotype. Further characterization of our newly generated Arid5b


^emQ522*


mice, and comparison with gene-trap models can reveal truncation specific deregulated mechanisms that could further pinpoint disease mechanisms specific for truncating variants that locate at the beginning of exon 10, like Arid5b


^emQ522*


. Challenging the immune system of these mice is a logical future approach to assess their immunological vulnerabilities.

ARID5B also plays a role during chondrogenesis by promoting the production of collagen type II in a manner dependent on SOX9 and involving the activation of SIRT1


³⁸


. This suggests that ARID5B contributes to extracellular matrix (ECM) formation and may support the maintenance of tissue integrity. Although homozygous mice are smaller than their heterozygous counterparts, subtle alterations in ECM formation or phenotypes manifesting only under specific challenges remain to be investigated. Such mechanisms could help explain the variably expressed systemic manifestations—such as hydronephrosis or immunological abnormalities—which may arise not only from direct immune cell perturbations but also indirectly through impaired barrier function and altered cell adhesion secondary to collagen dysregulation.

In conclusion, we have shown that truncating variants, including frameshift variants in exon 9, leading to stop-gains that cluster in the first quarter of the disproportionally long exon 10 of

ARID5B

, cause a neurodevelopment syndrome with a spectrum of phenotypes, including kidney issues, hypotonia and feeding problems, most frequently mild ID and with DD, speech and language delay/disorder and behavioral issues. A broad or bulbous nose, often with a short and low nasal bridge was observed. Other distinctive features included ptosis, radial hand deviations and foot/toe abnormalities. There is also a possible susceptibility to episodes of recurrent infections, often without a known cause, of the urinary tract, upper airways and inner ear, and two rare cases of CNS neuroinflammation, with one individual poorly responding to therapy being treated palliatively at the time of writing. Two infants presented with persistent pulmonary hypertension, leading to the death of individual 19 at six months of age. Truncating mutations escape NMD and showed cellular localization defects that we attribute to the loss or impairment of a C-terminal nuclear localization signal/nuclear export domain. Mice with truncating mutations in

ARID5B

show developmental and behavioral abnormalities. Missense variants in critical domains like the BRIGHT domain and truncating variants in the distal part of exon 10 very likely affect neurodevelopment, but their consequences may require individual assessment due to variable deleteriousness and partially unique phenotypes.

Data availability

The published article includes all datasets generated or analyzed during this study. Raw imaging data, constructs, and plasmid maps will be provided upon request. R code for statistics in

figures 4C

,

4G

, and

5G

, is available at

https://github.com/Hjvanheesbeen/ARID5B_Cohort_Statistics

.

Acknowledgements

Part of this work was performed under the Care4Rare Canada Consortium funded by Genome Canada and the Ontario Genomics Institute (OGI-147), the Canadian Institutes of Health Research, Ontario Research Fund, Genome Alberta, Genome British Columbia, Genome Quebec, and Children’s Hospital of Eastern Ontario Foundation.

Genetic testing of individuals with the c.1419del (p.Asn434LysfsTer45) and the c.1365dupA (p.Glu456ArgfsTer31) was performed within the framework of the GAD (

Génétique des Anomalies du Développement

) collection and approved by the appropriate institutional review board of Dijon University Hospital (DC2011-1332). The work was supported by grants from Dijon University Hospital, the ISITE-BFC (PIA ANR), the European Union through the FEDER programs (PERSONALISE), and the Burgundy-Franche-Compté regional council (INTEGRA). The sequencing platform at the CNRGH was supported by the France Génomique National infrastructure, funded as part of the “Investissements d’Avenir” program, managed by the Agence Nationale pour la Recherche (contract ANR-10-INBS-09). The whole genome sequencing performed at the CNRGH was funded by the Laboratory of Excellence GENMED (Medical Genomics) Grant No. ANR-10-LABX-0013, managed by the National Research Agency (ANR) as part of the Investment for the Future program.

The CRISPR-Cas9 images (thumbnail) (image “Cas9 gene”) was reproduced from the original work by the author (2024) under the Creative Commons Attribution-ShareAlike 4.0 International License. No modifications were made. The inner ear illustration (thumbnail) was provided by MedicalGraphics.de and is used under a Creative Commons Attribution-NoDerivatives 4.0 International License (CC BY-ND 4.0). No modifications were made to the original image. The drawing of the lungs Medical illustration (thumbnail) by Patrick J. Lynch and C. Carl Jaffe, Yale University School of Medicine, licensed under CC BY 2.5.

Megaureter illustration courtesy (thumbnail) of the Children’s Hospital of Philadelphia, used under public domain. The image “Cartoon brain illustration” (thumbnail) was obtained from the public domain collection on Rawpixeland is free to use without restriction under the CC0 Public Domain Dedication.

Additional information

Ethics Declaration

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of CHU Sainte-Justine (protocol code 015-853:4072 and date of approval 26 May 2015). Written informed consent from the participants’ legal guardian/next of kin was provided in accordance with the national legislation and the institutional requirements for the provided images to be published and donated blood to be used for scientific purposes.

Author Contributions

H.v.H wrote the manuscript and designed the experiments. H.v.H, N.R, A.G, A.P, and J.C. performed the experiments. H.v.H., N.R. and A.P. analyzed the experiments, P.C. & H.v.H. supervised the project, H.v.H. and P.C. recruited the cohort, and K.B. edited the manuscript. All other authors contributed by reviewing the manuscript and providing clinical information and/or exome analyses.

Additional files

Supplemental Information

Table S1

References

• 
  
  1.
  
  
  
  
  1. 
     Deák G
     
  
  
  2. 
     Cook AG
     
  
  
  
  
  2022
  
  
  
  Missense Variants Reveal Functional Insights Into the Human ARID Family of Gene Regulators
  
  
  
  
  Journal of Molecular Biology
  
  
  434
  
  :167529
  
  
  
  https://doi.org/10.1016/J.JMB.2022.167529
  
  
  Google Scholar
  
  
  


• 
  
  2.
  
  
  
  
  1. 
     Kosho T
     
  
  
  2. 
     Okamoto N
     
  
  
  3. 
     Collaborators CSSI
     
  
  
  
  
  2014
  
  
  
  Genotype-phenotype correlation of Coffin-Siris syndrome caused by mutations in SMARCB1, SMARCA4, SMARCE1, and AR-ID1A
  
  
  
  
  American Journal of Medical Genetics Part C: Seminars in Medical Genetics
  
  
  166
  
  :262–275
  
  
  
  https://doi.org/10.1002/ajmg.c.31407
  
  
  Google Scholar
  
  
  


• 
  
  3.
  
  
  
  
  1. 
     Vals MA
     
  
  
  2. 
     Õiglane-Shlik E
     
  
  
  3. 
     Nõukas M
     
  
  
  4. 
     et al.
     
  
  
  
  
  2014
  
  
  
  Coffin–Siris Syndrome with obesity, macro-cephaly, hepatomegaly and hyperinsulinism caused by a mutation in the ARID1B gene
  
  
  
  
  Eur J Hum Genet
  
  
  22
  
  :1327–1329
  
  
  
  https://doi.org/10.1038/ejhg.2014.25
  
  
  Google Scholar
  
  
  


• 
  
  4.
  
  
  
  
  1. 
     Schrier Vergano SA
     
  
  
  
  
  2024
  
  
  
  ARID2, a milder cause of Coffin-Siris Syndrome? Broadening the phenotype with 17 additional individuals
  
  
  
  
  American Journal of Medical Genetics Part A
  
  
  194
  
  :e63540
  
  
  
  https://doi.org/10.1002/ajmg.a.63540
  
  
  Google Scholar
  
  
  


• 
  
  5.
  
  
  
  
  1. 
     Lelieveld SH
     
  
  
  2. 
     Reijnders MRF
     
  
  
  3. 
     Pfundt R
     
  
  
  4. 
     et al.
     
  
  
  
  
  2016
  
  
  
  Meta-analysis of 2,104 trios provides sup-port for 10 new genes for intellectual disability
  
  
  
  
  Nature Neuroscience
  
  
  19
  
  :1194–1196
  
  
  
  https://doi.org/10.1038/nn.4352
  
  
  Google Scholar
  
  
  


• 
  
  6.
  
  
  
  
  1. 
     Gudlaugsdottir S
     
  
  
  2. 
     Boswell DR
     
  
  
  3. 
     Wood GR
     
  
  
  4. 
     Ma J
     
  
  
  
  
  2007
  
  
  
  Exon size distribution and the origin of introns
  
  
  
  
  Genetica
  
  
  131
  
  :299–306
  
  
  
  https://doi.org/10.1007/S10709-007-9139-4
  
  
  Google Scholar
  
  
  


• 
  
  7.
  
  
  
  
  1. 
     Lindeboom RGH
     
  
  
  2. 
     Supek F
     
  
  
  3. 
     Lehner B
     
  
  
  
  
  2016
  
  
  
  The rules and impact of nonsense-mediated mRNA decay in human cancers
  
  
  
  
  Nature genetics
  
  
  48
  
  :1112
  
  
  
  https://doi.org/10.1038/NG.3664
  
  
  Google Scholar
  
  
  


• 
  
  8.
  
  
  
  
  1. 
     Philippakis AA
     
  
  
  2. 
     Azzariti DR
     
  
  
  3. 
     Beltran S
     
  
  
  4. 
     et al.
     
  
  
  
  
  2015
  
  
  
  The Matchmaker Exchange: A Platform for Rare Disease Gene Discovery
  
  
  
  
  Human mutation
  
  
  36
  
  :915
  
  
  
  https://doi.org/10.1002/HUMU.22858
  
  
  Google Scholar
  
  
  


• 
  
  9.
  
  
  
  
  1. 
     Schindelin J
     
  
  
  2. 
     Arganda-Carreras I
     
  
  
  3. 
     Frise E
     
  
  
  4. 
     et al.
     
  
  
  
  
  2012
  
  
  
  Fiji: an open-source platform for biologi-cal-image analysis
  
  
  
  
  Nature Methods
  
  
  9
  
  :676–682
  
  
  
  https://doi.org/10.1038/nmeth.2019
  
  
  Google Scholar
  
  
  


• 
  
  10.
  
  
  
  
  1. 
     Sobreira N
     
  
  
  2. 
     Schiettecatte F
     
  
  
  3. 
     Valle D
     
  
  
  4. 
     Hamosh A
     
  
  
  
  
  2015
  
  
  
  GeneMatcher: A Matching Tool for Connecting Investigators with an Interest in the Same Gene
  
  
  
  
  Human Mutation
  
  
  36
  
  :928–930
  
  
  
  https://doi.org/10.1002/HUMU.22844
  
  
  Google Scholar
  
  
  


• 
  
  11.
  
  
  
  
  1. 
     Karczewski KJ
     
  
  
  2. 
     Francioli LC
     
  
  
  3. 
     Tiao G
     
  
  
  4. 
     et al.
     
  
  
  
  
  2020
  
  
  
  The mutational constraint spectrum quanti-fied from variation in 141,456 humans
  
  
  
  
  Nature
  
  
  581
  
  :434–443
  
  
  
  https://doi.org/10.1038/s41586-020-2308-7
  
  
  Google Scholar
  
  
  


• 
  
  12.
  
  
  
  
  1. 
     von Ehr J
     
  
  
  2. 
     Oberstrass L
     
  
  
  3. 
     Yazgan E
     
  
  
  4. 
     et al.
     
  
  
  
  
  2024
  
  
  
  Arid5a uses disordered extensions of its core ARID domain for distinct DNA- and RNA-recognition and gene regulation
  
  
  
  
  Journal of Biological Chemistry
  
  
  300
  
  :107457
  
  
  
  https://doi.org/10.1016/j.jbc.2024.107457
  
  
  Google Scholar
  
  
  


• 
  
  13.
  
  
  
  
  1. 
     Satterstrom FK
     
  
  
  2. 
     Kosmicki JA
     
  
  
  3. 
     Wang J
     
  
  
  4. 
     et al.
     
  
  
  
  
  2020
  
  
  
  Large-Scale Exome Sequencing Study Im-plicates Both Developmental and Functional Changes in the Neurobiology of Autism
  
  
  
  
  Cell
  
  
  180
  
  :568
  
  
  
  https://doi.org/10.1016/J.CELL.2019.12.036
  
  
  Google Scholar
  
  
  


• 
  
  14.
  
  
  
  
  1. 
     Krumm N
     
  
  
  2. 
     Turner TN
     
  
  
  3. 
     Baker C
     
  
  
  4. 
     et al.
     
  
  
  
  
  2015
  
  
  
  Excess of rare, inherited truncating mutations in autism
  
  
  
  
  Nat Genet
  
  
  47
  
  :582–588
  
  
  
  https://doi.org/10.1038/ng.3303
  
  
  Google Scholar
  
  
  


• 
  
  15.
  
  
  
  
  1. 
     Hori K
     
  
  
  2. 
     Nagai T
     
  
  
  3. 
     Shan W
     
  
  
  4. 
     et al.
     
  
  
  
  
  2015
  
  
  
  Heterozygous Disruption of Autism susceptibility candi-date 2 Causes Impaired Emotional Control and Cognitive Memory
  
  
  
  
  PLOS One
  
  
  10
  
  :e0145979
  
  
  
  https://doi.org/10.1371/journal.pone.0145979
  
  
  Google Scholar
  
  
  


• 
  
  16.
  
  
  
  
  1. 
     Saré RM
     
  
  
  2. 
     Levine M
     
  
  
  3. 
     Smith CB
     
  
  
  
  
  2016
  
  
  
  Behavioral Phenotype of Fmr1 Knock-Out Mice during Active Phase in an Altered Light/Dark Cycle
  
  
  
  
  eNeuro
  
  
  3
  
  :ENEURO.0035-16.2016
  
  
  
  https://doi.org/10.1523/ENEURO.0035-16.2016
  
  
  Google Scholar
  
  
  


• 
  
  17.
  
  
  
  
  1. 
     Contesse T
     
  
  
  2. 
     Ayrault M
     
  
  
  3. 
     Mantegazza M
     
  
  
  4. 
     Studer M
     
  
  
  5. 
     Deschaux O
     
  
  
  
  
  2019
  
  
  
  Hyperactive and anxio-lytic-like behaviors result from loss of COUP-TFI/Nr2f1 in the mouse cortex. Genes
  
  
  
  
  Brain and Behavior
  
  
  18
  
  :e12556
  
  
  
  https://doi.org/10.1111/gbb.12556
  
  
  Google Scholar
  
  
  


• 
  
  18.
  
  
  
  
  1. 
     Ba AN Nguyen
     
  
  
  2. 
     Pogoutse A
     
  
  
  3. 
     Provart N
     
  
  
  4. 
     Moses AM
     
  
  
  
  
  2009
  
  
  
  NLStradamus: a simple Hidden Markov Model for nuclear localization signal prediction
  
  
  
  
  BMC Bioinformatics
  
  
  10
  
  :202
  
  
  
  https://doi.org/10.1186/1471-2105-10-202
  
  
  Google Scholar
  
  
  


• 
  
  19.
  
  
  
  
  1. 
     Sigrist CJA
     
  
  
  2. 
     De Castro E
     
  
  
  3. 
     Cerutti L
     
  
  
  4. 
     et al.
     
  
  
  
  
  2013
  
  
  
  New and continuing developments at PRO-SITE
  
  
  
  
  Nucleic acids research
  
  
  41
  
  
  
  
  https://doi.org/10.1093/NAR/GKS1067
  
  
  Google Scholar
  
  
  


• 
  
  20.
  
  
  
  
  1. 
     Kosugi S
     
  
  
  2. 
     Hasebe M
     
  
  
  3. 
     Tomita M
     
  
  
  4. 
     Yanagawa H
     
  
  
  
  
  2009
  
  
  
  Systematic identification of cell cycle-dependent yeast nucleocytoplasmic shuttling proteins by prediction of composite motifs
  
  
  
  
  Proceedings of the National Academy of Sciences of the United States of America
  
  
  106
  
  :10171–10176
  
  
  
  https://doi.org/10.1073/PNAS.0900604106
  
  
  Google Scholar
  
  
  


• 
  
  21.
  
  
  
  
  1. 
     Pagni M
     
  
  
  2. 
     Ioannidis V
     
  
  
  3. 
     Cerutti L
     
  
  
  4. 
     et al.
     
  
  
  
  
  2007
  
  
  
  MyHits: improvements to an interactive resource for analyzing protein sequences
  
  
  
  
  Nucleic acids research
  
  
  35
  
  
  
  
  https://doi.org/10.1093/NAR/GKM352
  
  
  Google Scholar
  
  
  


• 
  
  22.
  
  
  
  
  1. 
     Gudmundsson S
     
  
  
  2. 
     Singer-Berk M
     
  
  
  3. 
     Watts NA
     
  
  
  4. 
     et al.
     
  
  
  
  
  2022
  
  
  
  Variant interpretation using population databases: Lessons from gnomAD
  
  
  
  
  Human Mutation
  
  
  43
  
  :1012–1030
  
  
  
  https://doi.org/10.1002/humu.24309
  
  
  Google Scholar
  
  
  


• 
  
  23.
  
  
  
  
  1. 
     Martinez-Mayer J
     
  
  
  2. 
     Vishnopolska S
     
  
  
  3. 
     Perticarari C
     
  
  
  4. 
     et al.
     
  
  
  
  
  2024
  
  
  
  Exome Sequencing Has a High Diagnostic Rate in Sporadic Congenital Hypopituitarism and Reveals Novel Candidate Genes
  
  
  
  
  The Journal of Clinical Endocrinology & Metabolism
  
  
  109
  
  :3196–3210
  
  
  
  https://doi.org/10.1210/clinem/dgae320
  
  
  Google Scholar
  
  
  


• 
  
  24.
  
  
  
  
  1. 
     Cichocki F
     
  
  
  2. 
     Wu CY
     
  
  
  3. 
     Zhang B
     
  
  
  4. 
     et al.
     
  
  
  
  
  2018
  
  
  
  ARID5B regulates metabolic programming in hu-man adaptive NK cells
  
  
  
  
  The Journal of Experimental Medicine
  
  
  215
  
  :2379
  
  
  
  https://doi.org/10.1084/JEM.20172168
  
  
  Google Scholar
  
  
  


• 
  
  25.
  
  
  
  
  1. 
     Liu D
     
  
  
  2. 
     Zhang XX
     
  
  
  3. 
     Li MC
     
  
  
  4. 
     et al.
     
  
  
  
  
  2018
  
  
  
  C/EBPβ enhances platinum resistance of ovarian cancer cells by reprogramming H3K79 methylation
  
  
  
  
  Nature Communications
  
  
  9
  
  :1739
  
  
  
  https://doi.org/10.1038/s41467-018-03590-5
  
  
  Google Scholar
  
  
  


• 
  
  26.
  
  
  
  
  1. 
     Saare M
     
  
  
  2. 
     Tserel L
     
  
  
  3. 
     Haljasmägi L
     
  
  
  4. 
     et al.
     
  
  
  
  
  2020
  
  
  
  Monocytes present age-related changes in phospholipid concentration and decreased energy metabolism
  
  
  
  
  Aging cell
  
  
  19
  
  
  
  
  https://doi.org/10.1111/ACEL.13127
  
  
  Google Scholar
  
  
  


• 
  
  27.
  
  
  
  
  1. 
     Tan SH
     
  
  
  2. 
     Leong WZ
     
  
  
  3. 
     Ngoc PCT
     
  
  
  4. 
     et al.
     
  
  
  
  
  2019
  
  
  
  The enhancer RNA ARIEL activates the oncogenic transcriptional program in T-cell acute lymphoblastic leukemia
  
  
  
  
  Blood
  
  
  134
  
  :239
  
  
  
  https://doi.org/10.1182/BLOOD.2018874503
  
  
  Google Scholar
  
  
  


• 
  
  28.
  
  
  
  
  1. 
     Kent WJ
     
  
  
  2. 
     Sugnet CW
     
  
  
  3. 
     Furey TS
     
  
  
  4. 
     et al.
     
  
  
  
  
  2002
  
  
  
  The Human Genome Browser at UCSC
  
  
  
  
  Genome Research
  
  
  12
  
  :996–1006
  
  
  
  https://doi.org/10.1101/GR.229102
  
  
  Google Scholar
  
  
  


• 
  
  29.
  
  
  
  
  1. 
     Okada Y
     
  
  
  2. 
     Terao C
     
  
  
  3. 
     Ikari K
     
  
  
  4. 
     et al.
     
  
  
  
  
  2012
  
  
  
  Meta-analysis identifies nine new loci associated with rheumatoid arthritis in the Japanese population
  
  
  
  
  Nature Genetics
  
  
  44
  
  :511–516
  
  
  
  https://doi.org/10.1038/ng.2231
  
  
  Google Scholar
  
  
  


• 
  
  30.
  
  
  
  
  1. 
     Ge Z
     
  
  
  2. 
     Han Q
     
  
  
  3. 
     Gu Y
     
  
  
  4. 
     et al.
     
  
  
  
  
  2018
  
  
  
  Aberrant ARID5B expression and its association with Ikaros dysfunction in acute lymphoblastic leukemia
  
  
  
  
  Oncogenesis
  
  
  7
  
  :1–10
  
  
  
  https://doi.org/10.1038/s41389-018-0095-x
  
  
  Google Scholar
  
  
  


• 
  
  31.
  
  
  
  
  1. 
     Uckun FM
     
  
  
  2. 
     Ma H
     
  
  
  3. 
     Zhang J
     
  
  
  4. 
     et al.
     
  
  
  
  
  2012
  
  
  
  Serine phosphorylation by SYK is critical for nuclear localization and transcription factor function of Ikaros
  
  
  
  
  Proc Natl Acad Sci U S A
  
  
  109
  
  :18072–18077
  
  
  
  https://doi.org/10.1073/pnas.1209828109
  
  
  Google Scholar
  
  
  


• 
  
  32.
  
  
  
  
  1. 
     Lahoud MH
     
  
  
  2. 
     Ristevski S
     
  
  
  3. 
     Venter DJ
     
  
  
  4. 
     et al.
     
  
  
  
  
  2001
  
  
  
  Gene Targeting of Desrt, a Novel ARID Class DNA-Binding Protein, Causes Growth Retardation and Abnormal Development of Re-productive Organs
  
  
  
  
  Genome Research
  
  
  11
  
  :1327–1334
  
  
  
  https://doi.org/10.1101/GR.168801
  
  
  Google Scholar
  
  
  


• 
  
  33.
  
  
  
  
  1. 
     Whitson RH
     
  
  
  2. 
     Tsark W
     
  
  
  3. 
     Huang TH
     
  
  
  4. 
     Itakura K
     
  
  
  
  
  2003
  
  
  
  Neonatal mortality and leanness in mice lacking the ARID transcription factor Mrf-2
  
  
  
  
  Biochemical and Biophysical Research Communications
  
  
  312
  
  :997–1004
  
  
  
  https://doi.org/10.1016/j.bbrc.2003.11.026
  
  
  Google Scholar
  
  
  


• 
  
  34.
  
  
  
  
  1. 
     Whitson RH
     
  
  
  2. 
     Li SL
     
  
  
  3. 
     Zhang G
     
  
  
  4. 
     Larson GP
     
  
  
  5. 
     Itakura K
     
  
  
  
  
  2021
  
  
  
  Mice with Fabp4-Cre ablation of Arid5b are resistant to diet-induced obesity and hepatic steatosis
  
  
  
  
  Molecular and cellular endocrinology
  
  
  528
  
  
  
  
  https://doi.org/10.1016/J.MCE.2021.111246
  
  
  Google Scholar
  
  
  


• 
  
  35.
  
  
  
  
  1. 
     Yamakawa T
     
  
  
  2. 
     Sugimoto K
     
  
  
  3. 
     Whitson RH
     
  
  
  4. 
     Itakura K
     
  
  
  
  
  2010
  
  
  
  Modulator recognition factor-2 regu-lates triglyceride metabolism in adipocytes
  
  
  
  
  Biochemical and Biophysical Research Communications
  
  
  391
  
  :277–281
  
  
  
  https://doi.org/10.1016/j.bbrc.2009.11.049
  
  
  Google Scholar
  
  
  


• 
  
  36.
  
  
  
  
  1. 
     Claussnitzer M
     
  
  
  2. 
     Dankel SN
     
  
  
  3. 
     Kim KH
     
  
  
  4. 
     et al.
     
  
  
  
  
  2015
  
  
  
  FTO Obesity Variant Circuitry and Adipo-cyte Browning in Humans
  
  
  
  
  The New England journal of medicine
  
  
  373
  
  :895
  
  
  
  https://doi.org/10.1056/NEJMOA1502214
  
  
  Google Scholar
  
  
  


• 
  
  37.
  
  
  
  
  1. 
     Usui N
     
  
  
  2. 
     Berto S
     
  
  
  3. 
     Konishi A
     
  
  
  4. 
     et al.
     
  
  
  
  
  2021
  
  
  
  Zbtb16 regulates social cognitive behaviors and neo-cortical development
  
  
  
  
  Transl Psychiatry
  
  
  11
  
  :242
  
  
  
  https://doi.org/10.1038/s41398-021-01358-y
  
  
  Google Scholar
  
  
  


• 
  
  38.
  
  
  
  
  1. 
     Hata K
     
  
  
  2. 
     Takashima R
     
  
  
  3. 
     Amano K
     
  
  
  4. 
     et al.
     
  
  
  
  
  2013
  
  
  
  ARTICLE Arid5b facilitates chondrogenesis by recruiting the histone demethylase Phf2 to Sox9-regulated genes
  
  
  
  
  Nat Commun
  
  
  
  
  https://doi.org/10.1038/ncomms3850
  
  
  Google Scholar
  
  
  


• 
  
  39.
  
  
  
  
  1. 
     Colin E
     
  
  
  2. 
     Duffourd Y
     
  
  
  3. 
     Tisserant E
     
  
  
  4. 
     et al.
     
  
  
  
  
  2022
  
  
  
  OMIXCARE: OMICS technologies solved about 33% of the patients with heterogeneous rare neuro-developmental disorders and negative exome sequencing results and identified 13% additional candidate variants
  
  
  
  
  Frontiers in Cell and Developmental Biology
  
  
  10
  
  :1021785
  
  
  
  https://doi.org/10.3389/FCELL.2022.1021785
  
  
  Google Scholar
  
  
  


• 
  
  40.
  
  
  
  
  1. 
     Silk M
     
  
  
  2. 
     Petrovski S
     
  
  
  3. 
     Ascher DB
     
  
  
  
  
  2019
  
  
  
  MTR-Viewer: identifying regions within genes under purifying selection
  
  
  
  
  Nucleic Acids Research
  
  
  47
  
  :W121–W126
  
  
  
  https://doi.org/10.1093/NAR/GKZ457
  
  
  Google Scholar
  
  
  


• 
  
  41.
  
  
  
  
  1. 
     Baba A
     
  
  
  2. 
     Ohtake F
     
  
  
  3. 
     Okuno Y
     
  
  
  4. 
     Yokota
     
  
  
  5. 
     et al.
     
  
  
  
  
  2011
  
  
  
  PKA-dependent regulation of the histone ly-sine demethylase complex PHF2-ARID5B
  
  
  
  
  Nature cell biology
  
  
  13
  
  :668–675
  
  
  
  https://doi.org/10.1038/NCB2228
  
  
  Google Scholar
  
  
  

Article and author information

Author information

1. 
   

   
   Hendrikus J van Heesbeen
   

   
   
   
   CHU Sainte-Justine Research Center, Montreal, Canada
   
   
   
   
   ORCID iD:
   
   0000-0002-9128-3637
   
   
   
   
   
   
   • 
     
     For correspondence:
     
     erikvanheesbeen@gmail.com
     
   
   
   
   


2. 
   

   
   Nazim Rabouhi
   

   
   
   
   CHU Sainte-Justine Research Center, Montreal, Canada
   
   
   


3. 
   

   
   Aurélie Gouronc
   

   
   
   
   CHU Sainte-Justine Research Center, Montreal, Canada
   
   
   


4. 
   

   
   Angéline Preto
   

   
   
   
   CHU Sainte-Justine Research Center, Montreal, Canada
   
   
   


5. 
   

   
   Justine Rousseau
   

   
   
   
   CHU Sainte-Justine Research Center, Montreal, Canada
   
   
   


6. 
   

   
   Jade Charbonneau
   

   
   
   
   CHU Sainte-Justine Research Center, Montreal, Canada
   
   
   


7. 
   

   
   Antonio Vitobello
   

   
   
   
   Université Bourgogne Europe, CHU Dijon Bourgogne, Laboratoire de Génomique Médicale, FHU-TRANSLAD, Centre de recherche Translationnelle en Médecine moléculaire – Inserm UMR1231 équipe GAD, Dijon, France
   
   
   
   
   ORCID iD:
   
   0000-0003-3717-8374
   
   
   
   


8. 
   

   
   Ange-Line Bruel
   

   
   
   
   Université de Bourgogne, UMR1231 GAD, Inserm - Université de Bourgogne and Unité Fonctionnelle Innovation en Diagnostic génomique des maladies rares, FHU-TRANSLAD, CHU Dijon Bourgogne, Dijon, France
   
   
   


9. 
   

   
   Anne-Sophie Denommé-Pichon
   

   
   
   
   Université de Bourgogne, UMR1231 GAD, Inserm - Université de Bourgogne and Unité Fonctionnelle Innovation en Diagnostic génomique des maladies rares, FHU-TRANSLAD, CHU Dijon Bourgogne, Dijon, France
   
   
   
   
   ORCID iD:
   
   0000-0002-8986-8222
   
   
   
   


10. 
    

    
    Estelle Colin
    

    
    
    
    CHU Angers, Service de Génétique médicale and UMR CNRS 6214-INSERM 1083, Université d’Angers, Angers, France
    
    
    


11. 
    

    
    Bertrand Isidor
    

    
    
    
    Nantes Université, CHU de Nantes, Service de Génétique médicale, CNRS, INSERM, l’institut du thorax, Nantes, France
    
    
    


12. 
    

    
    Sophie Nambot
    

    
    
    
    Université de Bourgogne, Centre de Référence Maladies Rares “Anomalies du développement et syndromes malformatifs”, Centre de Génétique, FHUTRANSLAD et Institut GIMI, CHU Dijon Bourgogne, Dijon, France
    
    
    


13. 
    

    
    Desiree DeMille
    

    
    
    
    ARUP Institute for Clinical and Experimental Pathology, Salt Lake City, United States
    
    
    


14. 
    

    
    Pinar Bayrak-Toydemir
    

    
    
    
    ARUP Institute for Clinical and Experimental Pathology, Salt Lake City, United States
    
    
    


15. 
    

    
    Nicola Longo
    

    
    
    
    Division of Medical Genetics, University of Utah, Salt Lake City, United States
    
    
    


16. 
    

    
    Boris Keren
    

    
    
    
    Genetic Department, Pitié-Salpêtrière Hospital, AP-HP.Sorbonne University, Paris, France
    
    
    


17. 
    

    
    Alexandra Afenjar
    

    
    
    
    Genetic Department, Armand-Trousseau Hospital, AP-HP.Sorbonne University, Paris, France
    
    
    


18. 
    

    
    Jolien S Klein Wassink-Ruiter
    

    
    
    
    Department of Genetics, University of Groningen, University Medical Center Groningen, Groningen, Netherlands
    
    
    


19. 
    

    
    Ingrid PC Krapels
    

    
    
    
    Department of Clinical Genetics, Maastricht University Medical Center, Maastricht, Netherlands
    
    
    


20. 
    

    
    Hannah Titheradge
    

    
    
    
    University of Birmingham, Birmingham, United Kingdom, Clinical Genetics, Birmingham Women’s and Children’s NHS Foundation Trust, Birmingham, United Kingdom
    
    
    


21. 
    

    
    Gavin Ryan
    

    
    
    
    West Midlands Regional Genetics Laboratory, Birmingham women’s and children’s NHS hospital trust, Birmingham, United Kingdom
    
    
    


22. 
    

    
    Matias Wagner
    

    
    
    
    Institute of Human Genetics, School of Medicine and Health, Technical University of Munich, Munich, Germany
    
    
    


23. 
    

    
    Jill A Rosenfeld
    

    
    
    
    Baylor Genetics Laboratories, Baylor College of Medicine, Houston, United States
    
    
    
    
    ORCID iD:
    
    0000-0001-5664-7987
    
    
    
    


24. 
    

    
    David R Witt
    

    
    
    
    Genetics Department, Kaiser Permanente, San Jose, United States
    
    
    


25. 
    

    
    Anirudh Saronwala
    

    
    
    
    Medical Genetics and Genomics, Kaiser Permanente San Jose-Santa Clara Medical Center, San Jose, United States
    
    
    


26. 
    

    
    Yaping Yang
    

    
    
    
    AiLIfe Diagnostics, Pearland, United States
    
    
    
    
    ORCID iD:
    
    0000-0001-5608-8628
    
    
    
    


27. 
    

    
    Annick Rein-Rothschild
    

    
    
    
    Institute of Rare Diseases, Safra Children’s Hospital, Sheba Medical Center, Tel Hashomer, Israel
    
    
    


28. 
    

    
    Ortal Barel
    

    
    
    
    Genomics Unit, The Center for Cancer Research, Sheba Medical Center, Tel Hashomer, Israel
    
    
    


29. 
    

    
    Reena Jethva
    

    
    
    
    Center for Genetic and Genomic Medicine, Hackensack Meridian Health, Joseph M. Sanzari Children’s Hospital at HUMC, New Jersey, United States
    
    
    


30. 
    

    
    Saskia B Wortmann
    

    
    
    
    University Children’s Hospital Salzburg, Salzburg, Austria
    
    
    


31. 
    

    
    Katharina Diepold
    

    
    
    
    Clinic of Pediatrics and Neuropediatrics, Northeim, Germany
    
    
    


32. 
    

    
    Kevin Rostasy
    

    
    
    
    Department of Paediatric Neurology, Witten/Herdecke University, Children’s Hospital Datteln, Datteln, Germany
    
    
    


33. 
    

    
    Lola K Clarkson
    

    
    
    
    Greenwood Genetic Center, Columbia, United States
    
    
    


34. 
    

    
    Kathryn T Drazba
    

    
    
    
    Greenwood Genetic Center, Columbia, United States
    
    
    


35. 
    

    
    Raymond J Louie
    

    
    
    
    Greenwood Genetic Center, Greenwood, United States
    
    
    


36. 
    

    
    Himanshu Goel
    

    
    
    
    Hunter Genetics, The University of Newcastle, Newcastle, Australia
    
    
    


37. 
    

    
    Outi Kuismin
    

    
    
    
    Department of Clinical Genetics, Research Unit of Clinical Medicine, Medical Research Center Oulu, Oulu University Hospital and University of Oulu, Oulu, Finland
    
    
    


38. 
    

    
    Pekka Nokelainen
    

    
    
    
    Department of Pediatric Neurology, Tampere University Hospital, Tampere, Finland
    
    
    


39. 
    

    
    Jianling Ji
    

    
    
    
    Department of Pathology and Laboratory Medicine, Children’s Hospital of Los Angeles, and University of Southern California Keck School of Medicine, Los Angeles, United States
    
    
    


40. 
    

    
    Ashley Mills
    

    
    
    
    Department of Pathology and Laboratory Medicine, Children’s Hospital of Los Angeles, and University of Southern California Keck School of Medicine, Los Angeles, United States
    
    
    


41. 
    

    
    Matthew A Deardorff
    

    
    
    
    Department of Pathology and Laboratory Medicine, Children’s Hospital of Los Angeles, and University of Southern California Keck School of Medicine, Los Angeles, United States
    
    
    


42. 
    

    
    María Palomares-Bralo
    

    
    
    
    Department of Genetics, INGEMM, Institute of Medical and Molecular Genetics, Hospital Universitario La Paz, Madrid, Spain
    
    
    


43. 
    

    
    María-Ángeles Gómez-Cano
    

    
    
    
    Department of Genetics, INGEMM, Institute of Medical and Molecular Genetics, Hospital Universitario La Paz, Madrid, Spain
    
    
    


44. 
    

    
    Alberto Fernández-Jaén
    

    
    
    
    Neuropediatric Department, Hospital Universitario Quirónsalud, Universidad Europea de Madrid, Pozuelo de Alarcón, Spain
    
    
    
    
    ORCID iD:
    
    0000-0003-3306-9832
    
    
    
    


45. 
    

    
    Peter J Hulick
    

    
    
    
    Mark R. Neaman Center for Personalized Medicine, Endeavor Health, Evanston, United States
    
    
    


46. 
    

    
    Maureen Jacob
    

    
    
    
    Institute of Human Genetics, School of Medicine and Health, Technical University of Munich, Munich, Germany
    
    
    


47. 
    

    
    Benjamin Cogne
    

    
    
    
    Nantes Université, CHU de Nantes, Service de Génétique médicale, Nantes, France
    
    
    


48. 
    

    
    Kandamurugu Manickam
    

    
    
    
    Division of Genetic and Genomic Medicine, Nationwide Children’s Hospital/OSU College of Medicine, Columbus, United States
    
    
    
    
    ORCID iD:
    
    0000-0001-6269-7710
    
    
    
    


49. 
    

    
    Xueqi Wang
    

    
    
    
    Children’s Hospital of Eastern Ontario Research Institute, Ottawa, Canada
    
    
    


50. 
    

    
    Gail Graham
    

    
    
    
    Children’s Hospital of Eastern Ontario, Department of Genetics, Ottawa, Canada
    
    
    


51. 
    

    
    Bert Callewaert
    

    
    
    
    Center for Medical Genetics, Ghent University Hospital and Department of Biomolecular Medicine, Ghent University, Ghent, Belgium
    
    
    
    
    ORCID iD:
    
    0000-0002-9743-4205
    
    
    
    


52. 
    

    
    Mercedes Zoeteman
    

    
    
    
    Genomics Institute, MultiCare Health System, Tacoma, United States
    
    
    


53. 
    

    
    Michael L Raff
    

    
    
    
    Genomics Institute, MultiCare Health System, Tacoma, United States
    
    
    


54. 
    

    
    Marion Aubert Mucca
    

    
    
    
    Service de génétique médicale, CHU de Toulouse, Toulouse, France
    
    
    


55. 
    

    
    Médéric Jeanne
    

    
    
    
    Department of Genetics, Eurofins Biomnis, Lyon, France
    
    
    


56. 
    

    
    Grace Raines
    

    
    
    
    University of Alabama at Birmingham, Department of Genetics, Birmingham, United States
    
    
    


57. 
    

    
    Amy Crunk
    

    
    
    
    GeneDx, LLC, Gaithersburg, United States
    
    
    


58. 
    

    
    Sureni V Mullegama
    

    
    
    
    GeneDx, LLC, Gaithersburg, United States
    
    
    


59. 
    

    
    Taila Hartley
    

    
    
    
    Children’s Hospital of Eastern Ontario Research Institute, University of Ottawa, Ottawa, Canada
    
    
    


60. 
    

    
    Kristin Kernohan
    

    
    
    
    Children’s Hospital of Eastern Ontario, Newborn Screening Ontario, Ottawa, Canada
    
    
    


61. 
    

    
    Kym Boycott
    

    
    
    
    Children’s Hospital of Eastern Ontario Research Institute, University of Ottawa, Ottawa, Canada
    
    
    


62. 
    

    
    Philippe M Campeau
    

    
    
    
    CHU Sainte-Justine Research Center, Montreal, Canada
    
    
    
    
    ORCID iD:
    
    0000-0001-9713-7107
    
    
    
    
    
    
    • 
      
      For correspondence:
      
      p.campeau@umontreal.ca
      
    
    
    
    

Metrics

views

0

downloads

0

citations

0

Views, downloads and citations are aggregated across all versions of this paper published by eLife.