De novo variants in RYBP are associated with a severe neurodevelopmental disorder and congenital anomalies

Abstract

Purpose:

Polycomb group (PcG) proteins are key epigenetic regulators of gene transcription. Multiple neurodevelopmental disorders have been associated with pathogenic variants in genes encoding PcG proteins. RYBP is a core component of the non-canonical Polycomb Repressor Complex 1; however, its role in disease is unclear.

Methods:

Functional consequences of RYBP variants were assessed through in vitro cellular studies and in vivo Drosophila melanogaster studies.

Results:

We describe seven individuals with heterozygous de novo variants in RYBP and clinical findings including severe developmental delay, dysmorphisms and multiple congenital anomalies. We show that all the single nucleotide variants in RYBP localize to the N-terminal domain of the gene, which encodes the zinc finger domain and ubiquitin binding moiety. In vitro studies demonstrate that the RYBP c.132C>G p.(Cys44Trp) variant causes reduced protein expression but does not affect binding of YY1, RING1B or ubiquitin. In vivo overexpression studies in Drosophila melanogaster show a dramatic functional difference from human RYBP and variant forms affecting the C44 amino acid residue. DNA methylation studies suggest a possible episignature that may be associated with RYBP-related disorder.

Conclusion:

Heterozygous de novo variants in RYBP are associated with an identifiable syndromic neurodevelopmental disorder with multiple congenital anomalies.

Introduction

Polycomb group (PcG) proteins are master epigenetic regulators of gene expression during embryonic development¹. PcG proteins were first described in Drosophila melanogaster as transcription repressors, and PcG complex gene variants resulted in homeotic transformation^2,3. PcG proteins are classified into two groups: PRC1 (Polycomb Repressor Complex 1) and PRC2 (Polycomb Repressor Complex 2), both of which play a crucial role in embryonic development, carcinogenesis, stem cell maintenance and X inactivation^4–8. Both PRC complexes are involved in histone modifications and transcriptional repression. While PRC1 mono-ubiquitinates H2A histones, the PRC2 complex is involved in methylation of H3 histone tails. The function of the PRC complexes is regulated by the specific core subunits and the accessory binding proteins. The activity of the PRC1 and PRC2 is linked, and multiple studies suggest that they affect each other’s chromatin occupancy^9,10. Pathogenic variants in many PcG complex encoding genes have been associated with different neurodevelopmental disorders. Examples include overgrowth syndromes like Weaver syndrome, Imagawa-Matsumoto syndrome, Cohen-Gibson syndrome and others like AUTS2-related autosomal dominant intellectual developmental disorder 26^11–15.

RYBP is a C2C2 zinc finger protein identified in a yeast two-hybrid screen for mammalian RING1A-interacting protein partners¹⁶. RYBP is a core component of the non-canonical (variant) Polycomb Repressor Complex 1 (nc-PRC1)^7,16. Nc-PRC1 differs from the canonical PRC1 (c-PRC1) by the inclusion of RYBP, which replaces the CBX proteins in the c-PRC1. The other core components of the nc-PRC1 include RING1A/B and PcG ring finger 1-6 proteins^10,17. RYBP interacts with the C-terminal domain of RING1A/B and binds directly to YY1 transcription factor, thus it was named RING1 and YY1 binding protein (RYBP)¹⁶.

RYBP protein undergoes ubiquitination¹⁸ through the ubiquitin binding site localized in its N-terminal zinc finger domain, and it can bind other ubiquitinated proteins. The RING1A/B binding domain is in the C-terminal part of RYBP, and its binding is independent of RYBP ubiquitination. RYBP increases RING1B mediated H2A histone mono-ubiquitination¹⁹. RYBP binds directly to the transcription factor YY1¹⁶ and mediates its binding to other transcription factors^20,21. Given the association of de novo pathogenic variants in YY1 and RING1B with syndromic intellectual disability^22,23, RYBP is a strong candidate gene for a similar syndromic phenotype.

In this study we describe seven individuals with heterozygous de novo variants in RYBP and show that RYBP (HGNC:10480) variants are associated with an identifiable syndromic neurodevelopmental disorder with multiple congenital anomalies.

Materials and Methods

Study population

Clinical evaluation of the participants enrolled in this study was performed in the different recruiting institutes. Written informed consent was obtained from all the families in the study according to IRB regulation, including consent for publication of patient photographs. Proband 1 was enrolled in the Undiagnosed Diseases Network (UDN) clinical site at Baylor College of Medicine after clinical genetic evaluation did not yield a diagnosis. The participating collaborators were connected through GeneMatcher²⁴ and UDN.

Massively parallel sequencing and analysis

Exome sequencing (ES) was performed for proband 1 as described previously²⁵. Codified Genomics software was used for variant interpretation. Genome sequencing (GS) was performed for proband 1 as described previously²⁶.

Exome sequencing and variant analysis was performed for proband 2 as described previously²⁷.

Exome sequencing was performed for proband 3 using a SureSelect Human All Exon Kit 60 Mb, V6 (Agilent, Santa Clara, California) for enrichment and an Illumina NovaSeq6000 or Illumina HiSeq4000 system (Illumina, San Diego, California) for sequencing. Reads were aligned to the human reference genome build GRCh37 with BWA v.0.7.8. SNVs and small insertions and deletions were detected using SAMtools²⁸ v.0.1.19. Copy number variations (CNVs) were detected with ExomeDepth and Pindel²⁹. Variants were analysed in the in-house exome variant analysis database (EVAdb) using: 1), a recessive filter for homozygous and compound heterozygous variants with a minor allele frequency (MAF), according to in-house database with over 20 000 exomes, < 1%, 2) a filter for X chromosomal variants with a MAF < 0.1% and 3) a filter for de novo variants with a MAF < 0.01%. A phenotype-based search was conducted by performing an OMIM full term search using the three most characteristic phenotypic traits to establish a gene list. The filter queries variants with a MAF < 0.1%. In addition, CNVs with a MAF < 0.01% and mtDNA variants with a MAF < 1% were assessed. Identified variants were classified according to the American College of Medical Genetics and Genomics (ACMG) guidelines³⁰.

Exome sequencing was performed for proband 4 on DNA extracted from blood using SureSelect All Human Exome v5 (Agilent, Santa Clara, CA) according to manufacturer’s instruction. Enriched library was paired-end sequenced (2x100 bp) on HiSeq 1500 (Illumina, San Diego, California). Data analysis and variant prioritization were performed as previously described³¹.

Trio genome sequencing was performed and analyzed for proband 5 as previously described³².

A chromosomal microarray was performed on the DNA of the proband 6 and his parents with an Illumina OmniExpress-24 v1.2 array using the Infinium HTS protocol according to the manufacturer’s recommendations. CNV analysis was carried out with the Illumina GenomeStudio 2.0 genotyping module and the cnvPartition 3.1.6 plugin.

Exome sequencing was performed for proband 7. Libraries were prepared with Illumina’s Truseq DNA Sample Preparation Kit (Illumina, San Diego, California) and exome enrichment (Truseq exome enrichment kit; Illumina, San Diego, California), following the manufacturer’s protocol. Sequencing was done by 100-bp paired-end sequencing on a HiSeq2500 instrument (Illumina, San Diego, California). Reads were aligned to the human reference genome build (GRCh37) using Burrows-Wheeler transform alignment (BWAv.0.7.5). Next, PCR duplicates were removed using Picard v1.92, and base quality recalibration, indel realignment, single nucleotide variant (SNV) and indel discovery were performed using the Genome analysis toolkit (GATK v2.5-2). Lastly, data were filtered against dbSNP137³³, 1000 Genomes, and an in-house exome SNV database and then annotated with SnpEff³⁴ 3.2a.

RNA sequencing and analysis

RNA from whole blood and skin fibroblasts was quantified and processed using a stranded, polyA-tailed kit (Illumina, San Diego, California) before being multiplexed and subjected to 150 bp paired-end sequencing at the BCM Laboratory for Translational Genomics, with approximately 30–50 million reads generated per sample. The sequencing data were processed with a pipeline adapted from one developed by the GTEx Consortium³⁵.

Transgenic fly generation

Plasmid constructs with human cDNAs of reference and variant RYBP cDNA under UAS promoter were created by gateway cloning method, and transgenic flies were generated.

Fly husbandry

All flies were grown in a temperature- and humidity-controlled incubator at either 25°C or as indicated in the text. All animals were maintained at 50% humidity on a 12-hour light/dark cycle. Flies were allowed to feed on a standard fly food medium consisting of water, yeast, soy flour, cornmeal, agar, corn syrup and propionic acid. Collection of animals for experiments was performed during daylight hours.

Overexpression assay

Lethality and morphological phenotyping assays were performed by crossing UAS-RYBP transgenic fly lines to different GAL4 fly lines as indicated in the text, using 5-7 virgin females crossed to a similar number of males. Parents were transferred into a new vial every 3-5 days to collect multiple F1-progenies. Flies were collected after most pupae eclosed, and the total number of flies was scored based on the presence or absence of balancers. For the lethality assessment, a minimum of 100 flies were scored. Viability was calculated via assessment of the number of observed progenies compared to the number of expected progenies based on the Mendelian ratio. Animals were classified as lethal if the O/E ratio was less than 0.15, and semi-lethality is classified as an O/E ratio less than 0.8. Assessment of morphological phenotypes was only done for animals lacking balancers, and phenotypes were noted if they appeared in more than 70% of the progeny.

Cell culture and plasmids

293T cells were cultured in DMEM with 10% fetal calf serum and 1% L-glutamine, 1% penicillin/streptomycin. Prior to transfection, the cells were cultured to about 70% confluency. For the transfection we used the expression plasmids: PCMV6 Myc-DDK RYBP (Origene), His₆-Ub expression plasmid (gift from Dr. Dae-Sik Lim³⁶), pcDNA3.1 HA-YY1 (Addgene). PCMV6 Myc-DDK RYBP C44W expression plasmid was generated with QuickChange Lightning Site-Directed mutagenesis kit (Agilent) per manufacturer protocol.

Coimmunoprecipitation assay

293T cells were transiently transfected with Myc-RYBP vector and HA-YY1 using Lipofectamine 3000. After 48 hours, the cells were lysed in lysis buffer (50 mM Tris HCl, pH 8, 150 mM NaCl, 0.1% NP-40, 1 mM EDTA and proteinase inhibitor cocktail), vortexed, and then incubated on ice for 20 minutes. After centrifugation at 13,000 g for 20 minutes at 4°C, the supernatant was incubated with anti-HA antibody (InvivoGen) or normal IgG (Santa Cruz Biotechnology Inc.) and bound to protein G/A beads (Roche). After washing the immunoprecipitated samples 3 times with lysis buffer, proteins were separated by SDS-PAGE, and RYBP was detected with anti-Myc antibody (Santa Cruz Biotechnology Inc.).

Ubiquitination assay

293T cells were transiently transfected with Myc-RYBP vector and His₆-Ub plasmid. After 48 hours, the His-tagged ubiquitinated proteins were pulled down with Ni-NTA agarose beads as previously described³⁷. Following Ni-NTA pulldown, proteins were separated by SDS-PAGE and RYBP was probed with anti-Myc antibody.

Variant nomenclature

The genomic coordinates of the RYBP variants were annotated using hg37, NM_012234.6 (which is identical to NM_012234.7 coding sequence). The coding coordinates were manually reviewed and corrected based on genome patch NW_011332691.1 to account for the incomplete coverage of the RYBP gene by hg19 and hg38 assemblies, which lack the first exon (encoding 10 amino acids) and the 5’ UTR (Table 2).

Table 2:

Genetic variants summary

Proband	1	2	3	4	5	6	7
Variant (hg19) NM_012234.6 (NM_012234.7 identical CDS)	3:72495673GCAGATG>CCAGATA; [c.132C>G; c.126C>T]; [p.Cys44Trp;p.Ser42=]	3:72495674C>G; c.131G>C; p.Cys44Ser	3:72495659C>A; c.146G>T; p.Gly49Val	3:72495713G>A; c.92C>T; p.Thr31Ile	3:72492452-72496261del 3.8 kb deletion	3:72385884-72899742del 513.8 kb deletion	3:72428401C>G; c.335+1G>C
NC_000003.11	[g.72495673G>C; g.72495679G>A]	g.72495674C>G	g.72495659C>A	g.72495713G>A	n/a	n/a	g.72428401C>G
NW_011332691.1	[g.120964G>C; g.120970G>A]	g.120965C>G	g.120950C>A	g.121004G>A	n/a	n/a	g.53828C>G
Inheritance	De novo	De novo	De novo	De novo	De novo	De novo	De novo
Frequency (gnomAD)	0a	0	0	0	n/a	n/a	0
ClinVar	0a	0	0	0	0	0	0
CADD	26.5a	29.2	31	28.7	n/a	n/a	33
PolyPhen	Probably damaginga	Probably damaging	Probably damaging	Probably damaging	n/a	n/a	n/a
SIFT	Damaginga	Damaging	Damaging	Damaging	n/a	n/a	n/a
AlphaMissense	1a	1	1	1	n/a	n/a	n/a
Conservation (GERP)	2.21a	5.08	5.08	4.92	n/a	n/a	5.76

^adenotes c.132C>G variant, CDS-coding sequence

Results

Clinical description and genetic evaluation

Proband 1 is a 20-year-old male, born after uneventful pregnancy to non-consanguineous parents. He was born with an atrial septal defect that closed spontaneously in childhood. He was diagnosed with global developmental delay from an early age, as well as growth failure that required G-tube feeding in childhood. At age 20 years he was non-verbal. He suffered chronic constipation from an early age and underwent surgery due to bowel malrotation. Other medical problems included bilateral mild to moderate conductive hearing loss, exotropia, myopic amblyopia, asthma, and skeletal abnormalities.

Physical examination at age 20 years showed a showed short stature (Z=−7) with low weight for age (Z=−4) and normal head circumference (Z=+0.8). Dysmorphic features included hypertelorism, downslanting palpebral fissures with thick eyebrows, prominent forehead, and small ears. He had a small mouth with everted upper lip and short philtrum, high arched palate, micrognathia and prominent maxilla. The teeth were widely spaced, and the upper incisors were prominent (Figure 1A). Skeletal findings noted on examination included mild scoliosis, mesomelia and brachydactyly. Hyperlaxity of finger joints was observed with limitation of forearm extension and supination as well as bowing of forearms.

Figure 1:

De novo variants in RYBP are associated with facial dysmorphism. A. Proband 1 at age 15y (left 2 panels), proband 2 at 9 months of age (middle upper panel), proband 3 at age 4 years (middle lower panel), proband 4 at age 9 years (right panel). Common dysmorphic features include a broad forehead, downslanting palpebral fissures, prominent nose, and small mouth with downturned corners. B. Sanger sequencing results for proband 1 showing heterozygous base changes in proband (upper panel noted with black arrows). Parents’ results are shown in the 2 lower panels. C. Exome sequencing (ES) showing heterozygous variants in proband 1. The lower panel shows the expression of the variants on RNA sequencing (RNAseq) in fibroblasts from proband 1.

Skeletal survey at age 16 years confirmed physical findings and showed shortened long bones with bowing of radius and ulna, and synostosis in the proximal region. In addition, Madelung deformity and mild kyphosis were demonstrated in the radiographs.

Proband 2 is a female born at full term with appropriate weight, head circumference and length for gestational age. She presented for evaluation at age 13 months with global developmental delay and failure to thrive that required G-tube feeding. On evaluation she was noted to have intestinal malrotation and cardiac anomalies including atrial septal defect and pulmonary stenosis. She required nighttime ventilation due to recurrent desaturations. MRI of spine showed a cyst of the filum terminalis. On physical examination at the age of 13 months, she was hypotonic and had no words or social smile. Height was normal for age (Z=−1), while weight was low (Z=−2.3). Dysmorphic features included microcephaly (Z= −3.7), almond shaped eyes with wide bridge of the nose, and a narrow mouth (Figure 1A). On follow up at age 3 years and 11 months, her growth improved with G-tube feeding and she was normocephalic (Z=−1.6). She walked independently, but due to muscular weakness and chronic respiratory failure she required nighttime ventilation. She was treated empirically with pyridostigmine bromide (Mestinon^®), a cholinesterase inhibitor, and parents reported some improvement. Developmentally, she was non-verbal and had global developmental delay.

Proband 3 is a 4-year-old male who was first evaluated at age 9 months due to failure to thrive, developmental delay and microcephaly. On physical examination hypotonia and hypospadias were noted at age 9 months. Echocardiogram showed an atrial septal defect. At age 4 years, he showed dysmorphism including downslanting palpebral fissures, broad forehead, and mouth with downturning corners (Figure 1A).

Proband 4 is a 9-year-old female born at full term with very low birth weight (Z=−2.8). Postnatally, she was noted to have patent foramen ovale, conductive hearing loss with middle ear hypoplasia, strabismus, hypothyroidism and left renal agenesis. During early childhood she had gastroesophageal reflux and recurrent respiratory infections with hyperventilation episodes. At the age of 9 years, she had significant global developmental delay with no speech or cry. On physical examination she was hypotonic with dysmorphic features including microcephaly, limited facial expression, high anterior hairline, small ears, protruding eyes, small mouth with short philtrum, and everted lips (Figure 1A). She had short stature (Z=−2.1), lumbar lordosis, asymmetry of limbs and brachydactyly.

Proband 5 is a 27-month-old male delivered at 37 weeks of gestation with a birth weight of 2325 grams (Z=1.9) and length of 47 cm (Z=−1.1). His clinical course is notable for several congenital anomalies including a congenital left diaphragmatic hernia, atrial septal defect, and cleft palate. In addition, he has dysmorphic features including external ear canal stenosis and short toes. He failed his newborn hearing screen, and on evaluation middle ear anomalies were found which included hypoplasia of the malleolar heads with epitympanic fixation and suspected ossicular fusion, oval window atresia and likely dehiscent facial nerve canals. His neonatal course was complicated by a grade I intraventricular hemorrhage. Subsequent brain MRI showed ventriculomegaly of the lateral and third ventricles. He has required long term ventilatory support and feeding through a gastric feeding tube. His most recent anthropometrics at 22 months demonstrated failure to thrive with a weight of 10.5 kg (Z=−2.1), height of 80 cm (Z=−2.8) and head circumference of 58 cm (Z=0). He was diagnosed with developmental delay and autism.

Proband 6 is a 30-year-old male born at full term with gestational age-appropriate measurements. At age 30 years he had obesity, macrocephaly (Z=+3.8), and short stature (Z=−2.3). He was also diagnosed with hypogonadotropic hypogonadism, hyperthyroidism, intellectual disability, and behavioral concerns. Brain MRI was normal. On examination, he had bilateral ptosis, high arched palate, simplified ears with anterior ear lobe creases, brachydactyly and large halluces.

Proband 7 is a 4-year-old male born at full term. At age 4 years he presented with global developmental delay, dysmorphisms, hypotonia, and short stature (Z=−2.5). Weight was appropriate for age (Z=−1.6), but he had a G-button due to feeding problems. Additional findings included gastroesophageal reflux, myopia with ptosis, hypospadias, camptodactyly, planovalgus, and scoliosis. Dysmorphic features included microcephaly (Z=−2.5), facial diplegia, and low set, posteriorly rotated ears.

A heterozygous de novo, suspected pathogenic, single nucleotide variant (SNV) or copy number variant (CNV) in RYBP was identified in all seven probands (Table 2). Probands 1 and 2 each harbored a nucleotide substitution at position c.132 of NM_012234.7, which is within codon 44 of the encoded protein (Proband 1 c.132C>G p.(Cys44Trp) referred to as C44W, Proband 2 c.131G>C p.(Cys44Ser) referred to as C44S). Both variants are predicted to be damaging in-silico. The CADD (Combined Annotation Dependent Depletion)³⁸ scores of these variants are above 25 suggesting a damaging effect, they are novel (not in gnomAD), and the affected residue is conserved. Proband 1 also harbored a synonymous variant c.126C>T p.(Ser42=) in cis with the C44W variant (Table 2, Figure 1B). The synonymous variant has a low CADD score of 14.4, appears in gnomAD with a frequency of 0.0004 and does not affect splicing according to RNA sequencing results (Figure 1C). Probands 3 and 4 had de novo missense variants at residues c.146G>T p.(Gly49Val) (NM_012234.7, referred to as G49V) and c.92C>T p.(Thr31Ile) (NM_012234.7, referred to as T31I) respectively. Both variants are novel (not in gnomAD), have a high CADD score and are in a conserved region of the gene (Table 2). Proband 7 had a splice site variant c.335+1G>C predicted to affect splicing (SpliceAI³⁹ score 1 for donor loss) and cause nonsense mediated mRNA decay (NMDEscPredictor⁴⁰). The residue is highly conserved, the variant is not found in the general population database (gnomAD v2.1.1) and CADD score is 33. Probands 5 and 6 harbored a de novo heterozygous copy number variant in RYBP. Proband 5 harbored a 3.8 kb deletion involving exon 1 and part of intron 1 of RYBP, while proband 6 had a 513.8 kb deletion including all RYBP and the neighboring gene SHQ1. Since only autosomal recessive variants in SHQ1 have been associated with a neurodevelopmental disorder with dystonia and seizures⁴¹, we propose that the heterozygous deletion of SHQ1 would not be associated with the clinical findings of this proband. The CNVs were not present in ClinGen⁴² database or DGV⁴³ control genomic variation database.

The clinical presentation of the probands was similar, involving congenital anomalies in multiple organ systems. Developmental delay was reported in all the probands, 3/7 had absent speech and microcephaly was reported in 4/7 cases. Five probands had hypotonia. 3/7 probands had cardiac anomalies, 5/7 had failure to thrive, and 4/7 required gastrostomy feeding. The skeletal findings of the probands were variable; 5 of 7 probands presented with short stature, and 3 out of 7 had brachydactyly (Table 1). Proband 1 showed severe skeletal abnormalities, thus proband clinical genome sequencing was performed, which did not identify pathogenic variants in known disease genes. Dysmorphic features included broad forehead, hypertelorism, downslanting palpebral fissures, prominent nose, small mouth with downturned corners and everted lips. Additional findings included ocular problems like myopia, astigmatism, strabismus, and ptosis. Hearing problems, genitourinary system problems and gastrointestinal problems were also noted but with less frequency (Table 2).

Table 1:

Clinical findings summary

Proband	1	2	3	4	5	6	7
Sex/age last evaluated	male/20y	female/13m	male/9m	female/9y	male/27m	male/30y	male/4y
Growth parameters at birth
Gestation (week, day)	39w	39w,3d	n/a	38w	37w	37w	38w
Weight (kg)	2.450	2.495	n/a	1.820	2.325	2.850	n/a
Length (cm)	45.5 (0.45 m)	48 (0.48 m)	n/a	43 (0.43 m)	47 (0.47 m)	45 (0.45 m)	n/a
Head circumference (cm)	33	33	n/a	n/a	n/a	33	n/a
Growth parameters at last visit
Weight (Z score)	− 4.8	−2.3	n/a	0.1	−2.1	+4.1	−1.6
Height (Z score)	− 5	−1	n/a	−2.1	−2.7	−2.3	−2.5
Head circumference (Z score)	+0.8	−3.7	n/a	n/a	0	+3.8	−2.5
Growth	failure to thrive	failure to thrive	failure to thrive	failure to thrive	failure to thrive	obesity	n/a
Neurodevelopment	severe intellectual disability, absent speech	developmental delay, no smile, absent speech	developmental delay	developmental delay, absent speech, no cry	developmental delay, autism	moderate intellectual disability, behavior problems	global developmental delay
ENT	conductive hearing loss, high arched palate, sleep apnea in childhood	-	n/a	conductive hearing loss with middle ear hypoplasia, sleep apnea	cleft palate, external and middle ear anomalies	high arched palate	-
Ocular	myopic astigmatism, amblyopia, nasolacrimal duct obstruction	n/a	n/a	strabismus	n/a	bilateral ptosis	myopia, bilateral ptosis
Cardiac	ASD and VSD	ASD II, pulmonary stenosis	ASD	PFO	ASD	-	PFO and PDA
Pulmonary	asthma in childhood	nighttime ventilation due recurrent desaturation	n/a	hyperventilation, recurrent respiratory infections	congenital diaphragmatic hernia, on ventilator	-	recurrent respiratory infections
GI	G-button up to 10 y, s/p intestinal malrotation, s/p intestinal obstruction	G-button, intestinal malrotation	n/a	GE reflux	G-tube dependent	-	GE reflux, dysphagia, G-button
Renal/GU	unilateral undescended testes s/p repair, hypospadias	-	hypospadias	left renal agenesis	-	-	hypospadias, undescended testis
Muscular	hypotonia	hypotonia	hypotonia	hypotonia	-	-	hypotonia
Endocrine	-	-	-	hypothyroidism, central obesity	-	hypogonadotropic hypogonadism, hyperthyroidism	-
Skeletal	short stature, mesomelic dysplasia, Madelung deformities, brachydactyly, radioulnar synostosis, advanced bone age in childhood	n/a	n/a	short stature, lumbar hyperlordosis, asymmetry of limbs, brachydactyly	short stature, finger contractures, dysplastic right 4th toe	short stature, brachydactyly, large halluces	short stature, camptodactyly, planovalgus feet, scoliosis
Craniofacial dysmorphism	hypertelorism, prominent forehead, downslanting palpebral fissures, bushy eyebrows, small mouth, large incisors, relative macrocephaly	wide bridge of the nose, almond-shaped eyes, narrow mouth, microcephaly	Broad forehead, downslanting palpebral fissures, mouth with downturning corners, microcephaly	limited facial expression, high anterior hairline, small ears, protruding eyes, small mouth with short philtrum, everted lips, microcephaly	n/a	simplified ears with anterior creases on lobes, macrocephaly	microcephaly, posteriorly rotated ears with low-set pinnae, facial diplegia
Brain MRI	normal	n/a	n/a	normal	lateral and 3rd ventriculomegaly, old grade I intra-ventricular hemorrhage in occipital horns, enlarged subarachnoid spaces more on left side	normal	n/a

ASD= atrial septal defect, GE-gastroesophageal, n/a= not available, PFO= patent foramen ovale, s/p= status post, VSD= ventricular septal defect, -= not found

RYBP is intolerant to loss-of-function variants as the pLI score⁴⁴ is 1 (gnomAD v4.1.0). In addition, the Z score for missense variants is 2.2 (gnomAD v4.1.0), suggesting that it is intolerant to missense variation. To investigate the effect of the missense variant at residue C44W on RYBP RNA expression, we performed RNA sequencing on fibroblasts derived from proband 1. The results showed expression of both alleles with no cryptic splice site generation (Figure 1C).

Overexpression of the human RYBP missense variants C44S and C44W in Drosophila cause a milder phenotype

RYBP is a highly conserved protein among different species, especially at the zinc finger binding domain, which is encoded by exon 1 of RYBP (Figure 2B). In Drosophila melanogaster, RYBP is moderately conserved with a DIOPT (DRSC Integrative Ortholog Prediction Tool) score⁴⁵ of 9/16. To investigate the functional effect of two of the human variants located in the same amino acid residue (C44W and C44S), we generated transgenic Drosophila flies carrying the human-RYBP cDNA under the Upstream Activated Sequences (UAS) promoter. We utilized the GAL4-UAS system to overexpress the two variants – hRYBP^C44W and hRYBP^C44S along with the human reference (wild type) hRYBP^Ref. We used Tubulin-GAL4, Actin-GAL4, and daughterless-GAL4 for ubiquitous expression and nubbin-GAL4 (wing-specific) and pannier-GAL4 (dorsal-midline area-specific) for tissue-specific expression (Table 3). Ubiquitous overexpression of human hRYBP^Ref caused lethality, but the hRYBP^C44W and hRYBP^C44S variant overexpressing flies were viable. Moreover, with the wing-specific nubbin-GAL4 overexpression study, it was observed that hRYBP^Ref displayed a wing margin defect with notches (100% of the F1 progeny flies), but the variant flies had normal wings (Figure 3A).

Figure 2:

RYBP protein domains and conservation between different species. A. Schematic depiction of RYBP protein domains and variants identified in the probands. Protein residues encoding the NZF domain are denoted in cyan blue. The zinc-coordinating cysteines are in red, and critical residues for ubiquitin binding are designated by stars. The RING1 and YY1 binding domain is denoted in yellow. B. Protein alignment (CLUSTAL O (1.2.4)) between human, mouse and Drosophila Melanogaster RYBP. The zinc-coordinating cysteines are noted in red.

Table 3:

Overexpression of human reference RYBP and RYBP C44 variants in Drosophila Melanogaster with different GAL4 lines

	UAS-hRYBPRef	UAS-hRYBPC44W	UAS-hRYBPC44S
Tubulin-GAL4	Lethal	Viable	Viable
Actin-GAL4	Lethal	Viable	Viable
daughterless-GAL4	Lethal	Viable	Viable
nubbin-GAL4	Wing notching phenotype	No phenotype	No phenotype
pannier-GAL4	Strong wing and dorsal-midline patterning phenotype	Mild wing and dorsal-midline patterning phenotype	Mild wing and dorsal-midline patterning phenotype

Figure 3:

Overexpression of human and variant RYBP protein in Drosophila melanogaster models. A. Wing-specific overexpression of human reference RYBP in nubbin GAL4 line causes a wing margin defect phenotype that is not seen with overexpression of the C44 RYBP variants. B. Overexpression of human reference and variant RYBP in the dorsal-midline-specific pannier GAL4. LacZ overexpressing flies serve as controls. Dashed rectangle indicates dorsal-midline region. Female flies on the left, male flies on the right. Flies overexpressing hRYBP^Ref show midline defect with ridge formation in the thoraco-abdominal region. There is increased midline abdominal pigmentation. Wings show significant blistering. The animals overexpressing the hRYBP^C44W or hRYBP^C44S variants show less significant midline ridges and less wing abnormalities.

Since RYBP is a core subunit of the nc-PRC1 and is involved in cell fate determination, we expressed it in the dorsal-midline region with pannier-GAL4. Pannier is involved in cell fate determination (shown in dashed area Figure 3B). When hRYBP^Ref was overexpressed in the dorsal-midline region we observed dramatic cell loss with some thoraco-abdominal midline ridges and an increase in the abdominal pigmentation along the midline area (Figure 3B). Also, the wings developed pigmented blisters all over the wing blade (Figure 3B–C). However, the overexpression of the hRYBP^C44W and hRYBP^C44S variants with pannier-GAL4 displayed a significantly milder phenotype (less dramatic midline ridges and less wing blistering) in comparison to the hRYBP^Ref (Figure 3B–C).

Ubiquitous expression of hRYBP^Ref with Tubulin, Actin, and daughterless GAL4s led to toxic effects that were not observed with expression of the variant forms of RYBP. However, the overexpression with pannier led us to observe the toxic effects of the proteins in the midline region along with the normal tissue in the same fly. These findings demonstrate that expression of the reference leads to dramatic phenotypes while the variants have either less dramatic (pannier-GAL4) or no phenotype.

The C44W RYBP variant is associated with changes in DNA methylation signature.

The non-canonical PRC1 that includes RYBP as a core component is involved in gene transcription repression by H2A histone mono-ubiquitination. It was suggested previously that canonical PRC1 binding to chromatin is PRC2-dependent in H3 trimethylated regions and that nc-PRC1 chromatin binding is independent of PRC2^9,10. Pathogenic variants in YY1 have been found to be associated with changes in genome wide methylation and recently the epigenetic signature was published⁴⁶. Since RYBP binds YY1, we assessed genome wide methylation for proband 1 with clinically available EpiSign test^47,48. The test results did not match the YY1 associated epigenetic signature (Figure 4A) or any other reported epigenetic signatures associated with known syndromes. In addition, the episignature also differed from the episignature associated with the PRC2 complex disorders (Figure 4B). However, the episignature showed an altered methylation pattern that partially matched X-linked syndromic intellectual disability Snyder-Robinson type (SMS gene), X-linked intellectual developmental disorder type 93 (RBMX gene) and CHARGE syndrome (CHD7 gene). Thus, the data together suggest that a distinct epigenetic signature may be associated with RYBP-related disorder.

Figure 4:

Methylation episignature related to RYBP. A. MDS (multidimensional scaling) clustering of control samples (green) and Gabriele-de Vries syndrome (GADEVS, blue). Proband 1’s sample is designated in red. B. MDS plots of control samples (green) and PRC2-associated disorder samples (blue). Proband 1 sample designated in red.

The RYBP C44W variant protein level is reduced in vitro

As we observed that the RYBP variants involving the C44 amino acid residue had less severe effect in the overexpression studies performed in the Drosophila flies, we performed in vitro studies in 293T cells to assess the protein levels of wild type RYBP and the C44W variant. We found that wild type (WT) RYBP was expressed at a higher level than the variant C44W (Fig.5). Although previously RNA sequencing on fibroblasts performed for proband 1 (Figure 1C) did not show reduced transcription levels of RYBP, the protein level of the variant.C44W RYBP was reduced in the cell studies. This data correlates with the findings from the fly model, suggesting that the variant protein effects are reduced possibly due to reduced protein expression.

Figure 5:

RYBP C44W variant protein level is decreased in vitro. Left panel: 293T cells were transfected with either wild type RYBP or C44W variant RYBP, and protein level assessed with anti-RYBP antibody. Tubulin levels are shown for loading control. Right panel: graph showing ratio between wild type RYBP protein and C44W variant protein level after normalization to tubulin. SD and mean shown in graph (n=4).

RYBP ubiquitination is not affected by the C44W variant localized in the ubiquitin binding domain

RYBP undergoes ubiquitination as part of its functional regulation and serves as an adaptor protein that affects target protein binding and protein stability^7,18,49. The ubiquitin binding domain is localized within the NZF (Npl4 zinc finger) type zinc finger domain¹⁸. The cysteine residue that is mutated in probands 1 and 2 lies within the ubiquitin binding region of the NZF type zinc finger (x(n)–W–x–C–x(2)–C–x(3)–N–x(6)–C–x(2)–C–x(n)), thus we tested whether the C44W variant could lead to abnormal ubiquitination of RYBP. 293T cells were co-transfected with His-tagged ubiquitin and wild type Myc-tagged RYBP (WT) or variant Myc-RYBP (C44W). The level of ubiquitinated RYBP was assessed after anti-ubiquitin immunoprecipitation (IP). However, we found that the WT RYBP and variant RYBP protein undergo ubiquitination at a similar level (Figure 6).

Figure 6:

RYBP protein ubiquitination is not affected by RYBP C44W variant. 293T cells were transfected with wild type or variant Myc-tagged RYBP alone or in combination with His-tagged ubiquitin. His-tagged ubiquitinated proteins were pulled down with anti-His, and ubiquitinated RYBP was identified with anti-Myc antibody. Input (IN) shown for each pull down. IP-immunoprecipitation.

RYBP interaction with YY1 and RING1B is not affected by the C44W variant at the zinc finger binding domain.

RYBP was identified through a two-hybrid screen for RING1A binding partners. Since RYBP protein shows homology to YAF2 (YY1-associated factor 2), Garcia et al. investigated whether it binds YY1 and showed that YY1 C-terminal portion is important for the binding of RYBP¹⁶. We evaluated whether the missense variant C44W affects the interaction between RYBP and YY1. We co-transfected 293T cells with HA-YY1 and either variant or wild type Myc-RYBP. We performed co-immunoprecipitation of RYBP and YY1 and found that both the wild type and variant RYBP (C44W) co-immunoprecipitated with YY1 (Figure 7). This finding suggests that the C44W variant in the RYBP ubiquitin binding domain does not affect YY1 binding. Furthermore, we assessed the binding of the C44W RYBP variant to RING1B and found that it does not affect RING1B binding in IP studies (data not shown).

Figure 7:

RYBP and YY1 protein interaction is not affected by RYBP C44W variant. Pull down of YY1 with wild type RYBP or variant RYBP in 293T cells. Immunoprecipitation performed with anti-HA antibody and western blot with anti-Myc antibody for RYBP. Inp-input, IgG- immunoglobulin G (control), α-HA-anti HA.

Discussion

PcG proteins are important epigenetic regulators in human development⁵⁰. These proteins form modular multi-protein complexes that were identified initially as transcription repressors. The complexity and function of these multi-protein complexes are far from fully deciphered, and human disorders associated with variants in the genes encoding members of these complexes can help to dissect further the importance of each subunit. In this study, we report that de novo single nucleotide variants and copy number variants in RYBP are associated with a syndromic neurodevelopmental disorder.

Pathogenic variants in genes encoding both major PRC complexes have been associated with human developmental disorders (See reviews^17,50). Pathogenic variants in EZH2 and SUZ12, which encode parts of the PRC2 complex, are associated with neurodevelopmental and overgrowth disorders (Weaver syndrome and Weaver-like syndrome, respectively)^11,12. Recently dominant and recessive pathogenic variants in EZH1 have been associated with a neurodevelopmental disorder^51,52. De novo pathogenic variants in the genes encoding PRC1 complex core protein RING1B (RNF2) and RING1A have been associated recently with intellectual disability disorders^22,53. Furthermore, pathogenic variants in numerous genes that are regulated by PcG complexes cause syndromic neurodevelopmental disorders. Examples of these include Rubinstein-Taybi, Bohring-Opitz, and Bainbridge-Ropers syndromes as well as others⁵⁰. YY1 haploinsufficiency was shown to cause an intellectual disability syndrome characterized by variable cognitive problems, dysmorphic features, congenital anomalies and feeding difficulties²³. Interestingly, pathogenic variants in genes encoding components of the PRC1 complex are typically not associated with overgrowth. Instead, growth problems and microcephaly are frequently reported^22,23,50,54.

The clinical findings of the individuals with de novo suspected pathogenic variants in RYBP are similar to the findings in individuals with YY1-related disorder (Gabriele-de Vries syndrome) and RING1B-related disorder (Luo-Schoch-Yamamoto syndrome). The similarities include developmental delay, feeding difficulties, growth problems, eye problems, skeletal findings, and craniofacial dysmorphism.

Individuals with RYBP-related disorder have global developmental delay from early childhood, and their speech development is significantly affected. Three of the seven reported individuals in our cohort had absent speech. Absent speech was reported in only one individual with YY1-related disorder, but delayed speech development was seen in all individuals. The severity of the neurodevelopmental delay in individuals with de novo RYBP variants could be associated with RYBP’s function in neuronal differentiation. Studies from mice embryonic stem cells showed that Rybp modulates neuronal proliferation and differentiation, and Notch signaling is repressed by Rybp deficiency^55,56. Rybp null embryonic stem cells formed less mature neurons, astrocytes and oligodendrocytes as well ^55,57. In mice, Rybp expression was shown to be co-localized with Ring1a in the developing mouse central nervous system at E9.0. Homozygous deletion of Rybp in mouse is embryonic lethal at early stages E5.5-E6⁵⁸.

Severe feeding difficulties and growth problems in early childhood are common findings in individuals with RYBP-related disorder. Five individuals in our cohort had failure to thrive, and four required G button or G-tube feeding early in childhood. Intestinal malrotation was found in two individuals. Most of the individuals with YY1-related disorder are also described as having failure to thrive, and some needed G-tube feeding^23,54. Proband 6 was reported with phenotypic findings similar to overgrowth, which included significant weight gain and macrocephaly, but since this proband was the only proband with a large copy number variant including the whole RYBP gene it is difficult to predict whether these findings could be seen in other probands with RYBP-related disorder or whether the deletion of regulatory regions could contribute to this phenotype.

Ocular problems like myopia, astigmatism, strabismus, amblyopia, and ptosis were found in four out of the seven individuals in our cohort. Ocular findings were also found frequently in YY1-related disorder. In chimeric Rybp knockout mouse embryos, coloboma and malformed lenses were found, suggesting that Rybp is important in eye development and function⁵⁹.

The characteristic craniofacial findings in RYBP-associated disorder include hypertelorism, broad forehead, downslanting palpebral fissures, prominent nose, and small mouth with downturned corners (Figure 1A). Similar facial features are observed in individuals with YY1 or RING1B pathogenic variants although with more variability^22,23,46,54.

Skeletal findings in individuals with RYBP-related disorder are variable including mesomelic limb shortening, short stature, scoliosis, limb asymmetry, camptodactyly and brachydactyly. The RING1B and YY1-associated skeletal findings include camptodactyly, scoliosis, limb asymmetry, joint laxity, and short stature^22,23,46,54. Previously, it was reported that ring1b is required for craniofacial development in zebrafish, and it has an important role in chondrocyte and bone differentiation⁶⁰. The study showed that ring1b-deficient zebrafish had almost complete absence of cranial cartilage, bone, and musculature. The cranial neural crest cells failed to undergo cartilage differentiation⁶⁰. Multiple studies showed that PRC2 proteins have an important role in regulation of skeletal development^61–63, but the role of the nc-PRC1 in chondrogenesis and osteogenesis is uncharacterized.

Congenital heart defects like atrial septal defects are observed frequently in RYBP-associated disorder while they are infrequent in individuals with YY1 pathogenic variants. In vitro studies previously showed that Rybp deficient embryonic stem cells do not differentiate normally to cardiomyocytes and that Rybp is important to the regulation of various cardiac-specific genes, thus cardiac malformations could be related to these cell type-specific functions^64,65.

The study cohort in this report includes individuals with missense variants, a splice variant and copy number variants. Mechanistically, missense variants can cause disease by either gain of function, neomorphic effect, dominant negative effect, or reduced activity (hypomorphic) while splice changes and deletion are often associated with loss of function (when mRNA nonsense mediated decay is present) and haploinsufficiency mechanisms. Previous reports on YY1 and RING1B-related disorders, suggested a haploinsufficiency mechanism although many of the reported variant were missense variants^22,23. To elucidate the potential mechanisms underlying RYBP-related disorder we performed in vitro and in vivo studies. To assess in vivo functional consequences of the variants at the C44 residue, we generated transgenic Drosophila melanogaster models, in which we utilized the versatility of the UAS/GAL4 system to overexpress the human reference and variant RYBP proteins. While ubiquitous overexpression of the reference protein was lethal, the overexpression of the missense variants was not. In wing-specific overexpression, the reference human protein caused significant wing anomalies that were not seen with the C44W or C44S variant overexpression. The overexpression of dRYBP has been previously shown to cause wing defects due to an increase in apoptosis⁶⁶. It is possible that the damage caused by our overexpression of hRYBP may be the result of the same pathway. On the other hand, overexpression of the reference RYBP in the dorsal midline-specific Drosophila fly model caused severe wing blistering and patterning defect that was significantly milder with the overexpression of the two C44 variant RYBPs (Figure 3). Together, the overexpression studies from Drosophila melanogaster suggest that either the C44 residue variants are functionally deficient, or protein expression is affected.

Previous literature suggested that RYBP protein folding is affected by binding partners and natively this protein does not have a well-defined secondary structure⁶⁷. The localization of RYBP missense variants in our cohort to the N-terminal zinc finger domain and ubiquitin binding domains suggested that these domains are potentially important in pathogenesis of the associated phenotypes (Figure 2). We assessed the expression of the C44W RYBP variant protein in 293T in comparison to wild type RYBP and observed reduced protein expression (Figure 5). Although our RNA sequencing data from proband 1 did not show reduced mRNA transcription, the expression of the variant protein in 293T cells was reduced. This could be explained by either reduced translation or increased protein degradation/protein instability. We further evaluated in vitro whether the missense variant C44W causes abnormal ubiquitination of RYBP, but we did not find significant differences in RYBP ubiquitination in comparison to the wild type of protein (Figure 6). YY1 and RING1B interaction was also not affected (Figure 7 and data not shown).

RYBP is a core element in the nc-PRC1, and it has an important role for the repressive function of the complex¹⁹. Epigenetic signatures are established early in development, and PcG complexes are vital to establishing the expression pattern of different genes during embryonic development. In recent years, with the development of newer technologies many different methylation episignatures were described related to PRC1 and PRC2 complexes^46,68,69. We hypothesized that variants in RYBP could affect the epigenetic signature related to the nc-PRC1 and assessed the methylation signature for proband 1 with the use of a clinically available EpiSign methylation array. We found that the proband’s methylation signature did not match the known methylation signature of patients with Gabriele-de Vries syndrome or the known PRC2-related episignature (Figure 4A–B). However, the sample showed an altered methylation pattern that partially matched X-linked syndromic intellectual disability Snyder-Robinson type, X-linked intellectual developmental disorder type 93 and CHARGE syndrome. These findings suggest a possible new methylation episignature that has not been identified yet and more probands with pathogenic variants in RYBP will be needed to establish the specific methylation signature.

In summary, in this study we describe seven individuals with de novo variants in RYBP who present with a novel, overlapping, syndromic neurodevelopmental disorder with identifiable dysmorphic features. Our experimental data suggests that changes in RYBP protein expression and secondary DNA methylation changes related to the nc-PRC1 repressive dysfunction could underlie the etiology of this new disorder.

List of Undiagnosed Disease Network workgroup:

Maria T. Acosta, David R. Adams, Ben Afzali, Ali Al-Beshri, Eric Allenspach, Aimee Allworth, Raquel L. Alvarez, Justin Alvey, Ashley Andrews, Euan A. Ashley, Carlos A. Bacino, Guney Bademci, Ashok Balasubramanyam, Dustin Baldridge, Erin Baldwin, Jim Bale, Elsa Balton , Michael Bamshad, Deborah Barbouth, Pinar Bayrak-Toydemir, Anita Beck, Alan H. Beggs, Edward Behrens, Gill Bejerano, Hugo J. Bellen, Jimmy Bennett, Jonathan A. Bernstein, Gerard T. Berry, Stephanie Bivona, Elizabeth Blue, John Bohnsack, Devon Bonner, Nicholas Borja, Lorenzo Botto, Lauren C. Briere, Elizabeth A. Burke, Lindsay C. Burrage, Manish J. Butte, Peter Byers, William E. Byrd, Kaitlin Callaway, John Carey, George Carvalho, Thomas Cassini, Sirisak Chanprasert, Hsiao-Tuan Chao, Ivan Chinn, Gary D. Clark, Terra R. Coakley, Laurel A. Cobban, Joy D. Cogan, Matthew Coggins, F. Sessions Cole, Brian Corner, Rosario I. Corona, William J. Craigen, Andrew B. Crouse, Vishnu Cuddapah, Precilla D’Souza, Hongzheng Dai, Nitsuh K. Dargie, Kahlen Darr, Surendra Dasari, Joie Davis, Margaret Delgado, Esteban C. Dell’Angelica, Katrina Dipple, Daniel Doherty, Naghmeh Dorrani, Jessica Douglas, Emilie D. Douine, Dawn Earl, Lisa T. Emrick, Christine M. Eng, Cecilia Esteves, Kimberly Ezell, Elizabeth L. Fieg, Paul G. Fisher, Brent L. Fogel, Jiayu Fu, William A. Gahl, Rebecca Ganetzky, Emily Glanton, Ian Glass, Page C. Goddard, Joanna M. Gonzalez, Andrea Gropman, Meghan C. Halley, Rizwan Hamid, Neal Hanchard, Kelly Hassey, Nichole Hayes, Frances High, Anne Hing, Fuki M. Hisama, Ingrid A. Holm, Jason Hom, Martha Horike-Pyne, Alden Huang, Yan Huang, Anna Hurst, Wendy Introne, Gail P. Jarvik, Suman Jayadev, Orpa Jean-Marie, Vaidehi Jobanputra, Oguz Kanca, Yigit Karasozen, Shamika Ketkar, Dana Kiley, Gonench Kilich, Eric Klee, Shilpa N. Kobren, Isaac S. Kohane, Jennefer N. Kohler, Bruce Korf, Susan Korrick, Deborah Krakow, Elijah Kravets, Seema R. Lalani, Christina Lam, Brendan C. Lanpher, Ian R. Lanza, Kumarie Latchman, Kimberly LeBlanc, Brendan H. Lee, Kathleen A. Leppig, Richard A. Lewis, Pengfei Liu, Nicola Longo, Joseph Loscalzo, Richard L. Maas, Ellen F. Macnamara, Calum A. MacRae, Valerie V. Maduro, AudreyStephannie Maghiro, Rachel Mahoney, May Christine V. Malicdan, Rong Mao, Ronit Marom, Gabor Marth, Beth A. Martin, Martin G. Martin, Julian A. Martínez-Agosto, Shruti Marwaha, Allyn McConkie-Rosell, Ashley McMinn, Matthew Might, Mohamad Mikati, Danny Miller, Ghayda Mirzaa, Breanna Mitchell , Paolo Moretti, Marie Morimoto, John J. Mulvihill, Lindsay Mulvihill , Mariko Nakano-Okuno, Stanley F. Nelson, Serena Neumann, Donna Novacic, Devin Oglesbee, James P. Orengo, Laura Pace, Stephen Pak, J. Carl Pallais, Neil H. Parker, LéShon Peart, Leoyklang Petcharet, John A. Phillips III, Filippo Pinto e Vairo, Jennifer E. Posey, Lorraine Potocki, Barbara N. Pusey Swerdzewski, Aaron Quinlan, Daniel J. Rader , Ramakrishnan Rajagopalan, Deepak A. Rao, Anna Raper, Wendy Raskind, Adriana Rebelo, Chloe M. Reuter, Lynette Rives, Lance H. Rodan, Martin Rodriguez, Jill A. Rosenfeld, Elizabeth Rosenthal, Francis Rossignol, Maura Ruzhnikov, Marla Sabaii, Jacinda B. Sampson, Timothy Schedl, Lisa Schimmenti , Kelly Schoch, Daryl A. Scott, Elaine Seto, Vandana Shashi, Emily Shelkowitz, Sam Sheppeard, Jimann Shin, Edwin K. Silverman, Giorgio Sirugo, Kathy Sisco, Tammi Skelton, Cara Skraban, Carson A. Smith, Kevin S. Smith, Lilianna Solnica-Krezel, Ben Solomon, Rebecca C. Spillmann, Andrew Stergachis, Joan M. Stoler, Kathleen Sullivan, Shamil R. Sunyaev, Shirley Sutton, David A. Sweetser, Virginia Sybert, Holly K. Tabor, Queenie Tan , Arjun Tarakad, Herman Taylor, Mustafa Tekin, Willa Thorson, Cynthia J. Tifft, Camilo Toro, Alyssa A. Tran, Rachel A. Ungar, Adeline Vanderver, Matt Velinder , Dave Viskochil, Tiphanie P. Vogel, Colleen E. Wahl, Melissa Walker, Nicole M. Walley, Jennifer Wambach, Michael F. Wangler, Patricia A. Ward, Daniel Wegner, Monika Weisz Hubshman, Mark Wener, Tara Wenger, Monte Westerfield, Matthew T. Wheeler, Jordan Whitlock, Lynne A. Wolfe, Heidi Wood, Kim Worley, Shinya Yamamoto, Zhe Zhang, Stephan Zuchner.

Data availability

The deidentified data supporting the current study is available from the corresponding author on request.