Comment on Gustavson syndrome is caused by an in-frame deletion in RBMX associated with potentially disturbed SH3 domain interactions

Abstract

Johansson et al. recently described the genetic diagnosis of a large family with Gusatvson syndrome. The pathogenic variant in this family is an in-frame deletion in RBMX, also known as HNRNPG. This work expands the definition of the HNRNP-Related Neurodevelopmental Disorders and provides insights into analyzing the related conditions.

Gustavson syndrome is a severe neurodevelopmental disorder (NDD) consisting of profound developmental delay/intellectual disability (DD/ID), microcephaly, structural brain anomalies, epilepsy, vision defects, hearing loss, cardiac defects, and early death [1]. To date, it has been described in one large family with X-linked inheritance. Johansson et al. recently described the genetic cause of Gustavson syndrome: an indel in RBMX in a likely SH3 domain. The deletion alters gene expression and may impair protein binding between SH3 domains.

RBMX encodes for heterogeneous nuclear ribonucleoprotein G (hnRNPG). The hnRNPs constitute a large gene/protein family that have critical shared, complementary, and compensatory roles in RNA processing [2, 3]. The gene family includes over thirty genes, including major hnRNPs that bind directly to RNA and minor hnRNP-like proteins that regulate major hnRNPs [3]. This gene family has been studied extensively and has members implicated neurodegenerative diseases, cancer, and more recently, NDDs.

The HNRNP-Related Neurodevelopmental Disorders (HNRNP-RNDDs) encompass a group of molecularly and clinically related conditions [4]. These disorders are characterized by DD/ID, delayed or absent speech and language development and/or apraxia, hypotonia, epilepsy, and behavioral differences. To date, the HNRNP-RNDDs consist of HNRNPC-, HNRNPH1- (MIM: 620083), HNRNPH2- (MIM: 300986), HNRNPR- (MIM: 620073), SYNCRIP/HNRNPQ-, and HNRNPU-RNDDs (MIM: 617391), as well as Au-Kline syndrome (MIM: 616580) caused by deleterious variants in HNRNPK [5–26]. Ongoing research is characterizing several candidate HNRNP-RNDDs.

Johansson et al. have now provided further evidence for the RMBX-RNDDs to be included in that list, as both Shashi-type ID (MIM: 300238) and now Gustavson syndrome (MIM: 309555) have been determined to be caused by specific RBMX variants [1, 27]. RBMX hnRNPG is known to be highly expressed during neurodevelopment and play a role regulation of several pre- and posttranscriptional processes.

Johansson et al. described the genetic diagnosis of a large family with X-linked Gustavson syndrome [1]. The family affected by Gustavson syndrome was originally described in 1993, highlighting their long diagnostic odyssey [28, 29]. Johansson et al. have shown that Gustavson syndrome is a result of a pathogenic in-frame deletion (p.Pro162del) in a likely novel SH3 domain that disrupts transcription and potentially impairs SH3 domain binding. Gustavson syndrome has phenotypic overlap with the other known HNRNP-RNDDs, except for the early death, making it one of the most severe HNRNP-RNDDs. The only other HNRNP gene on the X chromosome is HNRNPH2, in which primarily missense variants cause HNRNPH2-RNDD. Interestingly, HNRNPH2 escapes X-inactivation [8]. While primarily affecting females, males with HNRNPH2-RNDD have a more severe presentation consisting of growth delay, global DD/ID, motor delay, speech delay, ADHD, seizures, hypotonia, microcephaly, vision anomalies, and dysmorphic facies. Thus, hemizygosity of variants in the two X-linked HNRNPs results in a severe phenotype. Considering both males and females, RBMX-RNDDs phenotypes most closely relate to HNRNPC-RNDD (Pearson’s correlation = 0.80) and HNRNPH1-RNDD (Pearson’s correlation = 0.71), the latter of which is among the more severe of the disorders (Fig. 1 and Table 1). However, this is not true when looking at protein structure correlations, with RBMX having a unique RNA recognition motif. This further highlights that while the HNRNP-RNDDs that have more molecular similarities are share more clinical features, it is likely their individual functions contribute to phenotypic differences as well.

Fig. 1. Correlations between phenotypic features of the HNRNP-RNDDs.

Red indicates a higher correlation while yellow is a weaker correlation. p values are FDR corrected. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Table 1.

Overlapping phenotypes of Gustavson syndrome, Shashi-type ID, and the other HNRNP-RNDDs (HNRNPC-, HNRNPH1-, HNRNPH2-, Au-Kline syndrome (HNRNPK), HNRNPR-, SYNCRIP-, and HNRNPU-RNDDS.

Phenotype	Gustavson syndrome	Shashi-type ID (n = 8 for all)	p values (Bonferroni corrected)	Other HNRNP-RNDDs	p values (Bonferroni corrected)
IUGR	77.8% (n = 9)	0%	1.234e−03	3.3% (n = 275)	4.705e−11/NS
Short stature	88.9% (n = 9)	0%	0.001	30.9% (n = 268)	0.002/0.3
DD/ID	100% (n = 8, profound)	100% (moderate)	NS	96.2% (n = 393)	NS
Motor delay	100% (n = 2)	0%	NS	57.8% (n = 235)	NS/0.004
Speech delay	100% (n = 2)	0%	NS	73.4% (n = 241)	NS/0.0001
Brain imaging abnormalities	85.7% (n = 7)	N/A	N/A	41% (n = 222)	0.04
Seizures	80% (n = 10)	12.5%	0.05	61.3% (n = 310)	NS/0.02
Hypotonia	80% (n = 10)	0%	0.003	62.5% (n = 280)	NS/0.001
Hypertonia/Spasticity	40% (n = 10)	0%	NS	9.2% (n = 271)	0.03/NS
Microcephaly	90% (n = 10)	0%	1.23e−03	24% (n = 271)	9.84e−05/NS
Restricted joint movement	30% (n = 10)	0%	NS	N/A	N/A
Hand/feet anomalies	50% (n = 10)	0%	NS	38.1% (n = 265)	NS
Blindness/low vision/reduced visual acuity	70% (n = 10)	0%	1.21e−02	9.5% (n = 264)	4.4e−05/NS
Optic nerve abnormalities/optic atrophy/optic dysplasia	70% (n = 10)	0%	1.21e−02	3% (n = 265)	9.79e−08/NS
Hearing impairment	60% (severe) (n = 10)	50% (late onset)	NS	8.4% (n = 227)	0.0004/0.01
Dysmorphic facies	40% (n = 10)	87.5%	NS	69.3% (n = 267)	NS
Puffy eyelids	10% (n = 10)	87.5%	0.009	0% (n = 218)	NS/4.4e−12
Abnormality of the palpebral fissures	0% (n = 10)	87.5%	0.0008	20.5% (n = 220)	NS/0.0005
Bulbous tip of nose	0% (n = 10)	87.5%	0.0008	3.2% (n = 217)	NS/1.5e−08
Abnormal ear morphology	30% (n = 10)	87.5%	NS	27.2% (n = 232)	NS/0.003
Abnormality of the lip	0% (n = 10)	87.5%	0.0008	21.6% (n = 222)	NS/0.0006
Microretrognathia	30% (n = 10)	0%	NS	5.5% (n = 219)	NS
Large testicles	0% (n = 10)	87.5%	0.0008	0% (n = 204)	NS/6.9e−12
Immune abnormalities	40% (n = 10)	0%	0.0008	8.8% (n = 229)	NS/6.9e−12

Fisher’s exact test with Bonferroni correction was used to determine differences between Gustavson syndrome, Shashi-type ID, and the other HNRNP-RNDDs.

The findings in Gustavson syndrome, particularly a variant in a novel domain, may help elucidate the functional impact of variants in other HNRNP genes. Based on the same analysis applied by Johannssen et al. (using a PPI cutoff of 0.6 or greater), we find that hnRNPC (amino acids 131–134), hnRNPK (amino acids 272–274, 289–293, 309–314), hnRNPR (amino acids 512–514), hnRNPU (amino acids 163–165), hnRNPUL1 (amino acids 78–80, 695–714, 774–781), and hnRNPUL2 (amino acids 109–112) have putative SH3 domains. No variants within these amino acid ranges have been identified. However, variants may be identified in the future, as well as in other novel domains. It is important to keep such disease mechanisms in mind, particularly for indels and missense variants. There have been protein elongation variants similar to those found in Shashi-type ID for Au-Kline syndrome, HNRNPR-, SYNCRIP-, and HNRNPU-RNDDs.

While a handful of inherited pathogenic variants with potentially incomplete penetrance have been reported for the HNRNP-RNDDs, as well as one case of likely skewed X-inactivation in a mother, most of cases of HNRNP-RNDDs have been found to occur de novo [4, 30]. It is interesting that the RBMX variants appear to be localized to families and no known de novo variants have been described in the literature. While rare in large neurodevelopmental disorder cohorts, this is likely due to ascertainment bias, as many of the individuals with Gustavson syndrome die prematurely. Outside of the large Gustavson syndrome and Shashi-type ID families, there are two reported missense variants in RBMX in ClinVar (p.Ile34Thr [ClinVar Variation ID: 1683586] and p.Asp333Tyr [ClinVar Variation ID: 1184408]), one missense in the Simons Simplex Collection (p.Ser261Asn [13063.p1]) and one protein elongating variant in DECIPHER (p.Arg355Lysfs*8 [Patient ID: 488393]), which would be expected to result in Shashi-type ID. Three of the four individuals are males, with the other one not having sex reported. Two of the four have unknown inheritance, but one missense variant is de novo and one is maternally inherited. Limited clinical information is available, but the three individuals with missense variants have DD/ID and autism spectrum disorder. The individual with the p.Arg355Lysfs*8 variant has severe ID, absent speech, microcephaly, and blindness. The reported individuals with Gustavson syndrome have blindness and optic atrophy, but none of the reported Shashi-type ID individuals do, suggesting that this individual in DECIPHER has an overlapping phenotype between the two conditions. It will be interesting overtime to see if the overlap continues to be defined and the disorders are merged into one. This has happened with Okamoto syndrome and Au-Kline syndrome with HNRNPK variants, and the opposite (one disorder being split into two) has occurred with HNRNPH1- and HNRNPH2-RNDDs.

RBMX has shared expression during development with other HNRNPs, which are enriched in the striatum during early fetal development and the amygdala in early-mid fetal development (Fig. 2). RBMX is also enriched during early fetal development in the cortex and hippocampus, which may contribute to its severe phenotype. Johansson et al. nicely showed altered gene expression in neuronal cells (SH-SY5Y cells) with the Gustavson syndrome RBMX variant over expressed. While there were not many genes with significant changes in expression, this type of work, along with other transcriptomic studies, will be very relevant for comparisons between the HNRNP-RNDDs.

Fig. 2. HNRNP gene expression developing fetal cortex tissues.

Specific brain region enrichment as determined by SEA, showing enrichment of expression of the NDD HNRNPs in the early fetal striatum and early-mid fetal amygdala. RBMX is independently significantly expressed in the early fetal cortex and hippocampus. SEA availble at http://genetics.wustl.edu/jdlab/csea-tool-2/. Data from Gillentine et al., 2021.

In summary, Johansson et al. have further solidified the RBMX-RNDDs in the HNRNP-RNDDs family. Many of the HNRNP-RNDDs can be identified in large NDD cohort studies. However, RBMX-RNDDs show the importance of identifying pathogenic genetic changes from the patients. Even though only a small number of individuals have been reported, it is likely that the pathomechanisms described for the RBMX-RNDDs have relevance for the other related disorders.

Author contributions

MAG provided the conception, analysis, and writing of this manuscript.

Funding

There was no specific funding for this manuscript.

Competing interests

The author declares no competing interests.