GeneMatcher Aids in the Identification of a New Malformation Syndrome with Intellectual Disability, Unique Facial Dysmorphisms, and Skeletal and Connective Tissue Abnormalities Caused by De Novo Variants in HNRNPK

Abstract

We report a new syndrome due to loss of function variants in the heterogeneous nuclear ribonucleoprotein K gene (HNRNPK). We describe two probands, one with a de novo frameshift (NM_002140.3: c.953+1dup), and the other with a de novo splice donor site variant (NM_002140.3: c.257G>A). Both probands have intellectual disability, a shared unique craniofacial phenotype, and connective tissue and skeletal abnormalities. The identification of this syndrome was made possible by a new online tool, GeneMatcher, which facilitates connections between clinicians and researchers based on shared interest in candidate genes. This report demonstrates that new web based approaches can be effective in helping investigators solve exome sequencing projects, and also highlights the newer paradigm of “reverse phenotyping”, where characterization of syndromic features follows the identification of genetic variants.

Whole exome sequencing (WES) has rapidly become the primary method for the discovery of causative genes in rare disease. However, the ability to correctly identify responsible causative variants can be challenging, particularly with ultra-rare disorders that frequently arise de novo. While in silico, in vitro, cell based and animal model systems are able to provide some validation in linking a gene variant to a phenotype, finding another proband with a similar phenotype and the same causative gene is invaluable. The need for improved connections between clinicians and researchers who have interest in the same genes and/or phenotypes has become increasingly apparent as more variants are identified with next generation technologies, and has in part been addressed by several novel tools now available on the internet. Recently, Sobreira et al. have developed a web based tool called GeneMatcher [2015] (http://www.genematcher.org). Unlike other online tools such as PhenomeCentral, which match primarily based on phenotype, GeneMatcher facilitates connections between clinicians and researchers based on shared interest in a candidate gene. Herein we illustrate how a web based match tool has played an instrumental role in supporting HNRNPK (MIM# 600712) as the causative gene for a novel syndrome. We will also describe the distinct clinical features of the syndrome that is associated with HNRNPK haploinsufficiency.

HNRNPK encodes the heterogeneous nuclear ribonucleoprotein K (hnRNP K), a nucleic acid binding protein which has diverse roles in regulation of chromatin remodelling, transcription, RNA stability and splicing, translation, and signal transduction [Bomsztyk et al., 2004; Barboro et al., 2014]. Many of these roles have been implicated in human disease, but none have been identified yet as causative genes for congenital malformation syndromes. Here we describe two extensively investigated probands without a diagnosis who have been investigated by two separate research groups. WES independently identified de novo putative loss of function variants in HNRNPK in each proband. Although both groups were convinced that HNRNPK was the correct gene based on familial segregation and in-silico analyses, neither group was able to independently confirm the gene as causative. Ultimately, confirmation of pathogenicity was made through a GeneMatcher connection. Both probands with the loss of function HNRNPK variants have the same distinct clinical syndrome, consisting of intellectual disability associated with characteristic craniofacial features, skeletal, connective tissue, and cardiac anomalies. Additionally, two individuals with deletions of 9q21 encompassing HNRNPK have been recently reported in the literature with phenotypes that significantly overlap with our probands [Pua et al., 2014; Hancarova et al., 2015], lending further support that haploinsufficiency of HNRNPK leads to a congenital malformation syndrome associated with intellectual disability.

Proband 1 was born to a 33-year-old G3P2 mother and 33-year-old father after an uncomplicated healthy pregnancy. However, prenatal ultrasound identified a two vessel cord and a cardiac septal defect with pericardial effusion. Amniocentesis revealed a normal male karyotype. He was delivered via caesarean for breech presentation at 37 weeks gestation due to preterm labor, and had respiratory distress for 2-3 days. Birthweight was 3.03kg (50^th centile), length 47 cm (10-50^th centile), and head circumference 32.5 cm (10-50^th centile).

Dysmorphic features were observed at birth. These features included plagiocephaly, shallow orbits with proptotic eyes, abnormal ears, broad nasal root with a dimple at the nasal tip, poorly defined philtrum, excess nuchal skin, cryptorchidism, and a sacral dimple with a coccygeal appendage. He was hypotonic. Postnatal echocardiogram showed two small ventricular septal defects. There was feeding difficulty and failure to thrive in early infancy. At 6 months, he had a prominent metopic suture and bitemporal narrowing. He had surgery for sagittal suture craniosynostosis, and repair of a nasal sinus tract connecting to an inclusion cyst. There were persistent middle ear effusions during childhood, and he currently has mixed conductive and sensorineural hearing loss. At age 12, he had surgery for progressive scoliosis. Spine MRI showed multiple segmentation and fusion defects in the thoracic spine, and extra lumbar vertebral bodies (Figure 1D). Syrinx was observed at T7-T9 and at T12, with a low lying conus and a lipomyelomeningocele at the terminal thecal sac. Brain MRI was normal.

Figure 1. Clinical features of probands with HNRNPK loss of function variants.

Proband 1 (A-G) and Proband 2 (H-K).

A) Proband 1 at 13 years. B) Hands of proband 1, with fifth finger clinodactyly, prominent joints and shorter distal phalanges. C) High palate and bifid uvula in proband 1. D) T2 coronal MRI of proband 1 at age 12, demonstrating multiple vertebral segmentation defects and thoracolumbar scoliosis. H) Proband 2 at 11 years. E and I) Bifid tongues with prominent midline groove in proband 2 (I), and prominent midline groove of tongue in proband 1 (E). F and G) Proband 1 at ages 14 months and 8 years. J and K) Proband 2 at 1 year and 8 years. L) Individual described by Hancarova et al. (2015) at age 6 years. M) Representative western blot demonstrating reduced hn RNPK protein expression in proband 2 fibroblasts (P2) versus control fibroblasts (C1-3). N) Comparison of mean hn RNPK protein expression levels determined from 3 independent western blots, p = 0.008426.

Republished with permission.

He was assessed by Medical Genetics at 13 years of age in Calgary. Weight was 36 kg (10^th centile), height was 147 cm (above 10^th centile), and head circumference was 54.5 cm (just above 50^th centile). He appeared to have a dolichostenomelic body habitus. He had dolicocephaly, a long face, long and down slanting palpebral fissures with proptotic eyes, and his lateral eyebrows were broad. He had a broad prominent nasal bridge and hypoplastic alae nasi (Figure 1A). His philtrum was long, he had an open bite due to severe malocclusion, and micro and retrognathia. His palate was high arched and with a bifid uvula (Figure 1C). His tongue had a prominent midline groove (Figure 1E). His ears were low set with thick helices, and underdeveloped antihelix on the right side . His hands appeared slender with 5^th finger clinodactyly (Figure 1B) and he had planovalgus feet with broad halluces bilaterally, and crowding of toes. His chest was asymmetric due to scoliosis with widely spaced nipples, and he had a small midline abdominal hernia.

Global developmental delay was noticeable by 6 months of age. He walked independently at 4 years of age. Currently at 17 years of age, he uses sign language and has verbal speech that is difficult to understand to an untrained ear. He requires assistance with activities of daily living such as hygiene and dressing. He was recently assessed by the Weschler Nonverbal Scale of Ability (Canadian) and the Adaptive Behaviour Assessment System (ABAS-II) with full scale scores in the extremely low range at <0.1 percentile, consistent with moderate cognitive delay. However, he appears to socialize well at school.

On review of family history, he has two older siblings who are healthy with normal development. There is no history of intellectual disability, congenital anomalies, recurrent pregnancy loss or early death in the family.

Array CGH (60K) showed a maternally inherited deletion of 308kb at 8q12.3 overlapping a single gene, ASPH, which was thought to be non-contributory. Loeys-Dietz and Shprintzen-Goldberg syndromes were considered as possible diagnoses, however, sequencing for TGFBR1, TGFBR2, and SKI were negative.

Proband 2 was born to a healthy 28-year-old G2P1 mother and a 33-year-old father. The pregnancy was uncomplicated. Prenatal ultrasound at 18 weeks showed bilateral renal pelvis dilatation, ventriculomegaly, and nuchal thickening. Amniocentesis revealed a normal male karyotype, with incidental 9qh+. He was born by spontaneous vaginal delivery at over 38 weeks gestation, with no respiratory distress. Birthweight was 3460g (90^th centile), with a length of 54 cm (>95^th centile) and head circumference of 34.5 cm (75^th centile).

Multiple anomalies were recognized at birth, including dysmorphic facies with a broad nasal ridge and bifid nasal tip, increased nuchal skin, postaxial polydactyly, sacral dimple, right cryptorchidism, and hypotonia. Post natal echocardiogram showed a bicuspid aortic valve, mild branch pulmonic stenosis, and patent ductus arteriosus. Renal investigations showed hydronephrosis and vesicoureteral reflux.

He was reassessed at 2 years and again at 8, 9 and 11 years of age in Baltimore. Follow up echocardiogram revealed resolution of pulmonic stenosis and development of aortic root dilation. He developed osteoporosis in early childhood, and had transient hypothyroidism. He also has severe gastroesophageal reflux, gastrointestinal dysmotility, and requires a jejunostomy feeding tube. He has a neurogenic bladder and has a continent stoma. He has hypertension secondary to chronic kidney disease. He had asthma and nighttime hypoxia in early childhood. Hearing has been normal. Ophthalmology exam showed an optic nerve pit, hyperopia, and megalocornea, and he had corneal abrasions secondary to exposure from lagophthalmos. He had a ridged metopic suture and possible craniosynostosis. He developed mild scoliosis in childhood, and had developmental hip dysplasia, and later coxa valga. Spine MRI showed the presence of extra lumbar vertebrae. He has symptoms of dysautonomia. Brain MRI showed hypomyelination and prominent vasculature suggesting abnormal venous drainage.

On exam at 11 years of age, weight was 31.5 kg (50^th centile), with a length of 135.5 cm (10-25^th centile), and head circumference of 54.4 cm (75^th centile). Metopic suture and occipital shelf were prominent. His eyebrows were sparse laterally. He had long palpebral fissures and mild ptosis. He had a short nose with broad nasal bridge, notched alae nasi, and overhanging columnella. His mouth was held open and was downturned, palate was high with normal uvula, and tongue was tethered and bifid with a prominent midline groove (Figure 1H,I). His ears were underdeveloped. He had small inverted nipples and a supernumerary nipple. He had pectus excavatum. He was hyperextensible, with restricted extension in some proximal interphalangeal joints. His hands and feet were narrow, and he had caudal displacement of the third digits of his feet.

He has developmental delay and attention deficit hyperactivity disorder. In early childhood, he had frequent self-stimulatory behaviour. At age 11, he uses a device or computer to communicate, and has a few words, but has hundreds of signs. Psychometric testing has placed him at approximately a 4-5 year level for cognition. He is not toilet trained and requires assistance with walking.

On review of family history, he has an older brother with attention deficit disorder and autism with low normal range IQ who is otherwise non-dysmorphic and healthy. The family history is otherwise non-contributory.

Investigations have included a normal BAC array in 2007 and a normal SNP array CGH. Kabuki syndrome and TGFβ pathway disorders were considered, however, sequencing for KMT2D, TGFBR1 and TGFBR2 were normal. FLNB sequencing was also normal.

Families of both proband 1 and proband 2 have provided informed consent, and exome sequencing studies for both families have complied with the ethical regulations of the University of Calgary and Johns Hopkins University institutional review boards respectively. Exome sequencing of proband 1 and his parents was performed through the Care for Rare Canada consortium. The Agilent SureSelect 50 Mb capture kit (V3) was used for target enrichment. The Illumina HiSeq platform was used for sequencing, and read alignment, variant calling, and annotation were done as for previous FORGE and Care for Rare Canada projects [Bernier et al., 2012]. Variants were filtered for non-synonymous variants, rare variants not present in dbSNP, and for de novo variants not present in either parent. Two de novo variants were identified. One variant was a non-frameshift insertion in NEFH. The second variant was a frameshift insertion in HNRNPK, NM_002140.3:c.953+1dup, which adds a cytosine residue in between the +1 and +2 splice sites, and is predicted to alter gene expression either through nonsense-mediated decay or a frameshift that results in protein truncation, p.(Gly319Argfs*6). The HNRNPK variant was confirmed to be present in the proband and absent in the parents by Sanger sequencing. Variants have been submitted to ClinVar: SCV000223813 (c.257G>A) and SCV000223814 (c.953+1dup).

Exome sequencing for proband 2 and his parents was performed through the Baylor Hopkins Center for Mendelian Genomics (BHCMG). The Agilent SureSelect Human All Exon V4 51Mb Kit was used for target capture. The Illumina HiSeq2000 platform was used for sequencing, read alignment, and variant calling, using methodology that has been previously reported [Hoover-Fong et al., 2014]. We identified potential causal variants by standard filtering criteria as formerly described, followed by the use of the PhenoDB Variant Analysis Tool [Sobreira et al., 2015] to design the prioritization strategy. We prioritized rare functional variants that were de novo, homozygous, compound heterozygous or X-linked in the proband, and excluded variants with a Minor Allele Frequency (MAF) > 0.01 in dbSNP, the Exome Variant Server (release ESP6500SI-V2) or 1000 Genomes Project [Abecasis et al., 2012], and excluded all variants found in our in-house controls. In the de novo variant analysis, we identified a variant in HNRNPK, NM_002140.3:c.257G>A. While this sequence change is predicted to alter an amino acid, p.(Arg86His), it occurs in the last codon of exon 5 and is predicted to abolish the splice donor site. This variant was also confirmed to be absent in parents by Sanger sequencing.

There are no loss of function variants in HNRNPK in the ExAC database (Exome Aggregation Consortium (ExAC), Cambridge, MA (URL: http://exac.broadinstitute.org) (May 2015) nor in the EVS database (Exome Variant Server, NHLBI GO Exome Sequencing Project (ESP), Seattle, WA (URL: http://evs.gs.washington.edu/EVS/) (May 2015). No variants have been reported at either of the two residues that have been changed in these two probands, and no deletions involving this gene have been reported in DECIPHER.

Using western blot analysis we assessed protein expression of hn RNPK in proband 2 versus control fibroblasts (Supp. Methods). The hnRNPK protein was significantly decreased in Proband 2, 48.5% of control levels in fibroblasts (P<0.01, Welch's t-test, Figure 1M,N), supporting the predicted impact of the c.257G>A splice variant on HNRNPK gene expression.

There are 3 alternatively spliced isoforms of hnRNP K described in GenBank, and 4 isoforms have been reported in the literature and in UniProt [Kimura et al., 2010]. These isoforms either differ in the 5’UTR, have an alternative basic or acidic C-terminus end consisting of either ⁴⁵⁹SGKFF⁴⁶³ or ⁴⁵⁹ADVEGF⁴⁶⁴ respectively, or missing amino acid residues between 111 -134 due to alternative splicing of exon 8. The splice site variant in proband 2 (c.257G>A) at the junction of exon 6 and intron 6, and the frameshift truncating variant in proband 1 (c.953+1dup, p.(G319Rfs*6)) located at the 5’ end of exon 12, are both expected to disrupt all of these known isoforms based on their location.

HNRNPK was submitted as a candidate gene to Genematcher (https://genematcher.org) independently by research groups in Calgary and Baltimore, in an attempt to find additional probands with variants in HNRNPK. GeneMatcher is a web based program that was developed with support from the Baylor-Hopkins Center for Mendelian Genomics as part of the Centers for Mendelian Genomics. Candidate genes of interest are posted by investigators, and Genematcher connects investigators sharing a gene of interest via email. Genematcher was successful in facilitating a connection in this case. Subsequent sharing of proband characteristics and careful reappraisal of phenotype, including photographs, revealed that both probands possessed striking similarities, allowing identification of a novel syndrome and HNRNPK as the causative gene.

Here we report the discovery of a new distinct syndrome that is due to loss of function variants in the HNRNPK gene. Both probands share a common facial phenotype, characterized by long palpebral fissures, ptosis, a broad prominent nasal bridge, hypoplastic alae nasi, an open downturned mouth with a cupid's bow shaped upper vermilion and full lower lip, ears with underdeveloped and thick helices, and a unique tongue with a prominent median crease (Figure 1, Table 1). Both probands also share a connective tissue and skeletal phenotype with features that partially overlap with TGFβ pathway related disorders such as Loeys-Dietz or Shprintzen-Goldberg syndrome. Clinicians in both centers had pursued testing for TGFβ related syndromes (TGFBR1, TGFBR2 in both probands and SKI in proband 1), although neither proband was completely typical for these disorders [Loeys et al., 2005; Au et al., 2014]. Relevant features included a high palate (with bifid uvula in proband 1), scoliosis with vertebral segment defects and extra lumbar vertebrae, hip dysplasia, hyperextensibility, and craniosynostosis. Proband 2 also had aortic dilation and pectus excavatum. Other interesting features shared by these probands include decreased sweating and mild oligodontia (Table 1).

Table 1.

Clinical features of probands with haploinsufficiency of HNRNPK

	Proband 1 (male)	Proband 2 (male)	Pua et al. (female)	Hancarova et al. (female)
	HNRNPK c.953+1dup p.Gly319Argfs*6	HNRNPK c.257G>A p.Arg85His, splice site variant	Del 9q21.32-q21.33 (2.6Mb) (85,504,717-88,069,314) [HG19]	Del 9q21.32-q21.33 (2Mb) (86,595,071-88,357,495) [HG19)]
Growth parameters	At birth: Wt(50th%ile), L(10-50th%ile), HC(10-50th%ile). At 13y: Wt(10th %ile), Ht (>10th %ile), HC (>50th%ile)	At birth: Wt (90th%ile), L (>95th%ile), HC(75th%ile) At 9y: Wt (50th%ile), Ht(50-75th%ile), HC (75th%ile)	At birth: Wt(<50%ile), L(10%ile), HC(25%ile) At 6m: HC <3%ile	At birth: Wt (<50%ile), L (25%ile), HC(<10%ile) At 13y, all parameters <3%ile
Craniofacial	ridged metopic suture, dolicocephaly sagittal and lambdoid craniosynostosis, long face	ridged metopic suture, turricephaly long face	broad face low anterior hairline	narrow forehead low posterior hairline
Eyes	long downslanting palpebral fissures proptosis, ptosis broad lateral eyebrows	hyperopia long palpebral fissures, ptosis optic nerve pit, megalocornea, lagophthalmos, sparse lateral eyebrows	epicanthal folds	hypermetropia long downslanting palpebral fissures
Ears	underdeveloped helices hearing loss (conductive and sensorineural)	underdeveloped thick helices	low set, cupped right ear	large, low set ears
Nose	wide nasal ridge, hypoplastic alae nasi	wide nasal ridge, cleft of alae nasi	depressed nasal tip	wide nasal ridge
Mouth	open bite, downturned mouth, high palate, bifid uvula prominent midline groove of tongue missing molar	open, downturned mouth high palate, normal uvula prominent midline groove / bifid tongue missing molar and incisor	cleft soft palate	open bite, large downturned mouth
Chest	widely spaced nipples	pectus excavatum inverted nipples, supernumerary nipples	NR	NR
CVS/ Resp	2 small VSDs	bicuspid aortic valve aortic root dilation	large ASD respiratory difficulty	AVSD R atrial hemangioma
Gastrointestinal	constipation in early childhood	GERD, cyclic vomiting, constipation, GI dysmotility, G-tube fed
Genitourinary	cryptochordism	crytopchordism, VUR, neurogenic bladder hydronephrosis	uterine didelphys	VUR
Skeletal	hip dysplasia, scoliosis extra lumbar vertebrae and multiple vertebral segmentation defects, elbow contractures	hip dysplasia, scoliosis, extra lumbar vertebrae, hyperextensible	hip dysplasia, proximal upper extremity and distal lower extremity shortening	hip dysplasia
Hands/Feet	planovalgus feet, large hallux, crowded toes, decreased creases on feet	post axial polydactyly overlapping toes	hindfoot deformity	single palmar crease talipes equinovarus, overlapping toes
Skin	sacral dimple with coccygeal appendage, decreased sweating, intermittent facial rash	sacral dimple decreased sweating, intermittent rash	NR	NR
Neuro	hypotonia, hyporeflexia, high pain tolerance	hypotonia, hyporeflexia, high pain tolerance, migraine	hypotonia	hypotonia, later spasticity
Development	mild intellectual disability, walks independently, commnicates with many words, short phrases and uses signs and devices	mild-moderate intellectual disability, ADHD, walks with assistance, communicates with few words, many signs, and uses devices	severe delay	severe intellectual disability
MRI findings	normal brain syrinx T7-T9,T12, terminal lipomyelomeningocele	hypomyelination	thalamostriate vasculopathy, thin corpus callosum, hypomyelination	NR

Legend: NR (not reported). ASD (atrial septal defect), VSD (ventricular septal defect), AVSD (atrioventricular septal defect). VUR (vesicoureteral reflux), GERD (gastroesophageal reflux).

Two individuals have been reported in the literature with deletions of chromosome 9q21 that encompass the HNRNPK gene. Pua et al. describe a de novo 2.6 Mb deletion of 9q21.32q21.33 encompassing 12 genes, in a delayed hypotonic infant who died at 14 months of age. This infant had facial dysmorphisms, cleft of the soft palate, atrial septal defect and aortic coarctation, bilateral hip dislocation, hindfoot deformity, and delayed myelination on brain MRI [Pua et al., 2014]. More recently, Hancarova et al. reported a 13 year old female with a de novo 2 Mb deletion of 9q21.32q21.33 encompassing at least 8 genes, with 7 genes in the region of overlap with the infant described by Pua et al. (2014), with a very similar facial gestalt to the two probands described in this paper (Figure 1L). This individual also had atrial septal defect, congenital hip dysplasia, equinovarus foot deformity, severe delay and hypotonia that later developed into spastic quadraparesis [Hancarova et al., 2015]. Hancarova et al. had highlighted NTRK2 as a possible candidate causative gene for this microdeletion, given that the deletion described by Pua et al. also deleted NTRK2 and the role of NTRK2 in neurodevelopment. While some of the differences observed in these 9q21 microdeletion patients are likely attributed to haploinsufficiency of other genes within the region, there is considerable phenotypic overlap with these microdeletion patients and our two probands with HNRNPK putative loss of function variants, particularly the female described by Hancorava et al. [2015]. This lends additional support that haploinsufficiency of HNRNPK has a critical role in development.

HNRNPK encodes heterogeneous nuclear ribonucleoprotein K, and was initially identified as a component of hnRNP particles. hnRNP K is involved with RNA and/or single strand DNA binding through its three KH domains. HnRNP K also shuttles between the nucleus and the cytoplasm, and has a K-protein interactive domain that mediates interactions with various other proteins. HnRNP K likely acts as a docking platform to allow interaction of kinases and other signal transduction factors with nucleic acid-related cellular activities, and has roles in chromatin remodelling, transcription, RNA stability and splicing, translation, and signal transduction [Bomsztyk et al., 2004; Barboro et al., 2014]. The renin-angiotensin pathway appears to be regulated by hnRNP K at several levels, including hnRNP K regulation of renin mRNA stability [Skalweit et al., 2003], and post transcriptional control of collagen synthesis downstream of the angiotensin II type I receptor [Thiele et al., 2004]. More recently, hnRNP K has also been implicated in synaptic plasticity through its effects on ERK kinase cascade activation [Folci et al., 2014], and has a role in translational regulation of proteins involved in axonogenesis [Hutchins and Szaro, 2013]. The role of hnRNP K in multiple pathways may explain its effects on multiple organ systems in our probands. Its functions in angiotensin signalling and synaptic plasticity are particularly interesting given the connective tissue and neurodevelopmental features in these probands, as well as the hypertension in patient 2.

The confirmation of HNRNPK as the causative gene in both probands would not have been possible without the ability for the investigators to connect with each other. In proband 1, HNRNPK had been identified as a possible candidate gene for more than 2 years, and more traditional methods, such as presentation of the proband at conferences and other attempts at finding a phenotypic match for the proband, had not been successful. Finding cohorts of patients with a very specific phenotype is highly useful when attempting to identify an underlying genetic etiology, and has been a powerful and effective approach in many exome studies aimed at gene discovery. Some examples include the identification of EZH2 as the causative gene for Weaver syndrome, and identification of SRCAP for Floating Harbor syndrome, which both relied greatly on accurate phenotyping in a cohort of individuals with previously well recognized ‘classic’ dysmorphic syndromes [Gibson et al., 2012; Hood et al., 2012]. However, this strategy can be difficult if the genetic disease affecting the proband is very rare and/or not previously described, or if the phenotype is subtle or heterogeneous. Furthermore, online tools that depend on matching by phenotype, such as PhenomeCentral [Girdea et al., 2013], can be limited by clinician interpretation of what features are most significant, which may lead to discrepancies in descriptions of phenotypes that are actually similar, particularly if features are non-specific. This may affect the ability to match, even if candidate genes are incorporated into the PhenomeCentral match algorithm. GeneMatcher allows a different approach, where investigators can search for potential matches based on the gene without information about phenotype, circumventing some of the bias and limitations of looking for matches based primarily on descriptions of phenotype. Further analysis of phenotype after matching can subsequently be done in reverse to determine if the match is appropriate. However, it would be important to be aware of possible pitfalls from this approach as well. For example, certain genes may be implicated in multiple distinct phenotypes depending on the nature of the variant, and these scenarios may complicate gene matching and reverse phenotyping. Additionally, it is possible that phenotyping exclusively in reverse based on a candidate gene may lead to unintentional bias or a clinical assessment that is overly focused. Therefore, patients would ideally be well phenotyped initially prior to candidate gene identification, with further assessment of shared characteristics to be performed after a gene match is made.

Nevertheless, there is an imminent need for effective ways to elucidate unsolved exomes as increasing numbers of patients undergo whole exome sequencing, and as exome sequencing moves further into clinical practice. Therefore, the ability to make matches based on gene and/or phenotype to validate causative variants is becoming extremely important. We have demonstrated the value and utility of this new approach to matching through the identification of a new congenital malformation syndrome associated with intellectual disability due to haploinsufficiency of HNRNPK. Our probands have vertebral defects, connective tissue abnormalities, structural congenital heart defects, and genitourinary abnormalities. Finally, the facial gestalt is particularly distinct, characterized by long palpebral fissures, ptosis, a broad prominent nasal bridge, prominent underdeveloped ears, and an open downturned mouth with a prominent median crease to the tongue.