INTRODUCTION

Features of the chromosome 19q13.11 deletion syndrome include early growth deficiencies, developmental delay, distinctive facial features, aplasia cutis congenita (ACC), hip dysplasia, digital and limb anomalies including ectrodactyly, and other malformations.1,2,3,4,5,6,7,8 Deletions range in size from 1.37–11 Mb with a minimum overlapping region (MOR) of 324 kb, without clear genotype-phenotype correlation.3,4,6 UBA2 lies adjacent to the MOR and has been proposed to underlie key aspects of the deletion phenotype including ACC and ectrodactyly.1,2,3,5,6 Limited patient data and lack of an animal model have prevented establishing UBA2 as the causative gene.

UBA2 plays a key role in the post-translational modification of protein (SUMOylation) by the addition of SUMO1 (small ubiquitin-like modifier) protein. UBA2 forms a heterodimer with SAE1 (SUMO-Activating Enzyme Subunit 1) and binds with SUMO1 in an ATP-dependent manner.9,10,11 Unlike ubiquitination, SUMOylation does not only target proteins for degradation, but is also involved in cell cycle regulation, subcellular trafficking, signal transduction, stress responses, and chromatin structure dynamics. SUMOylation alters protein kinases and transcription factors to maintain transcriptional regulation of tissue-specific gene expression.12

In this study, we report 16 additional individuals from seven unrelated families with de novo and familial UBA2 sequence variants who have highly variable but overlapping clinical presentations. In silico modeling and a zebrafish uba2 nullizygous phenotype provide further functional evidence for the pathogenicity of UBA2 as the key gene underlying the chromosome 19q13.11 microdeletion syndrome.

MATERIALS AND METHODS

Subject enrollment and clinical evaluations

Each described patient was evaluated by a clinical geneticist. Written informed consent was obtained for exome sequencing either on a clinical or research basis. A written informed consent was also obtained from subjects to publish their photos. Genomic DNA was extracted from whole blood from affected probands and their biological parents for exome sequencing. See supplement for details.

Zebrafish modeling of the phenotypic effects of uba2 variants

All animal experiments were conducted in accordance with recommendations of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (protocol NEI-679). Adult AB (Tubingen) and ABTL (Tubingen long fin) zebrafish strains were raised and maintained according to standard protocols as described.13

Whole-mount in situ hybridization

Wild type (WT) zebrafish embryos at different developmental stages (5 somite, 24, 35, 48, 72 hpf [hours postfertilization]), 5 and 7 dpf (days postfertilization) were fixed in preparation for performing in situ hybridization. See Supplemental methods for details.

CRISPR/Cas9 uba2 knockout line generation

CRISPR/Cas9 method was used to generate uba2 knockout zebrafish lines. See Supplemental methods for details.

mRNA rescue

To evaluate the impact of human UBA2 variants on encoded protein products, we utilized uba2-mutant fish to perform rescue studies with capped full-length human WT and missense alleles in messenger RNA (mRNA) transcribed with the T7 mMESSAGE mMACHINE kit (Ambion).

Please see Supplement for other methodology details.

RESULTS

Clinical studies

The cohort was gathered through GeneDx, a clinical molecular laboratory, and GeneMatcher. Investigators independently ascertained families with related phenotypes and rare candidate variants. Table 1 and the Supplement contain additional clinical details.

Table 1 Clinical features in study subjects and previous publications.
Full size table

Family 1: Family 1 (Figs. 1 and 2) is comprised of an affected mother and her four offspring. Two children have ACC. By report, the maternal grandmother and great grandmother also have histories of ACC. Other ectodermal changes are variable including thin scalp hair, xerosis, and dental anomalies. The index case (IV-4, Fig. 1a, b) has unilateral ectrodactyly of the hand. All of the other affected examined individuals have more subtle digital variations including camptodactyly, syndactyly, clinodactyly, and diminished distal flexion creases of the fingers. All affected individuals share a high anterior hairline and mild frontal bossing, and several, including the proband (IV-4), have slightly downslanted palpebral fissures. All have had highly variable neurodevelopmental problems, ranging from hypotonia to autism spectrum disorder in two of the brothers. Hypotonia generally persisted throughout childhood. Affected individuals had early growth deficiencies that improved with age. See Supplement for other details. All affected individuals studied are heterozygous for a UBA2 frameshift variant: c.816_817delAT, p.Trp273Alafs*13.

Fig. 1: Clinical phenotypes and pedigrees associated with UBA2-related syndrome.
figure 1

(a) Family 1, III-2: A. prominent forehead, high hairline, discolored peg-shaped teeth with gap between upper incisors, cleft chin. B. Low-set ear with simple cartilaginous pattern. C. diminished distal flexion creases. D. Brachydactyly, mild 2–3 syndactyly, clinodactyly of the 4th toe. Family 1, IV-4: E. Prominent forehead, high hairline, cleft chin, mildly downslanted palpebral fissures. F. Aplasia cutis congenita (ACC). G. Repaired ectrodactyly, hypoplastic distal flexion creases. H. Ichthyosis. Family 3, I-1: I. tall forehead, hypertelorism, broad nasal root, mild micrognathia. J. ACC. Family 3, II-1: K. Tall forehead, hypertelorism, broad nasal root, thin upper lip, medial eyebrow flare. L. ACC. Family 3, II-2: M. Facial features. N. Supernumerary nipple (arrow). Family 3, III-1: O. Tall forehead, low-set ears, micrognathia. P. ACC. Family 6, II-1: Q. High forehead, hypertelorism, bilateral epicanthal folds. R, S. bilateral 2–3 finger syndactyly, camptodactyly T. bilateral ectrodactyly of the feet. (b) Affected individuals are shown as filled symbols. Genotypes are shown below each individual who was genotyped. (c) Percentages of different clinical features variably expressed in UBA2-affected individuals based on available data. Previously reported UBA2 patients are also included in the percentages.

Full size image
Fig. 2: UBA2 syndrome–associated variants and molecular modeling.
figure 2

(a) Schematic representation of the UBA2 gene. Exons are shown in brown color boxes; introns and 3’ and 5’ UTRs are in light gray and purple, respectively. Newly reported UBA2 missense and loss-of-function variants are shown in red and green, respectively, while blue is used to represent previously reported UBA2 variants. (b) Schematic representation of UBA2 protein domains. The UBA2 protein domain carrying catalytically active sites of ubiquitin-activating enzyme is shown in light green. This domain has putative active sites to bind ATP, substrate, and zinc with the last of five conserved cysteine residues playing an important role in ubiquitin thioester complex formation. UBA2-C (C-terminus) and UAE-UbL (ubiquitin-like) domains are shown in pink and blue, representing the C-terminus of UBA2 protein. UAE-UbL is structurally similar to ubiquitin and is involved in E1-SUMO-thioester transfer to E2 conjugation protein. The amino acid changes for the aforementioned variants are shown in the same color scheme as (b). (c) Amino acid sequence alignments of the human UBA2 protein across different species at each of the residues reported with missense variants. (d) Molecular modeling of human UBA2 protein. Secondary structure helix, strand and coil regions are shown in purple, yellow and cyan blue shades, respectively. Forest green color is used to show residues of interest in proteins with wild type (WT) and missense changes and the ATP molecule is shown in brick red color. Blue color shows regions of hydrogen bonding and light pink shows residues involved in hydrogen bond formation with residues of interest. The distances to nearby residues are shown by dashed yellow lines. Last panel: UBA2 and UBC9 are shown in hot pink and cyan blue color, respectively. As per molecular modeling predictions, p.Gly24Val: Glycine is flexible enough to maintain torsion angles and is buried in the protein core to maintain local secondary structure. p.Asn56Thr. Introducing a smaller but more hydrophobic residue at Asparagine 56 results in an empty space in the protein core and subsequent loss of hydrogen bonding with Asp53. p.Arg122Gly: The typical Arginine 122 residue is involved in hydrogen bonding with Asn118, Gly138 and Ser139. Replacement with Glycine is predicted to disrupt this array.

Full size image

Family 2: This family consists of three affected brothers (Fig. 1b: II-1, II-2, II-3); neither parent is affected. Parentage was genetically confirmed prior to exome sequencing. All affected individuals have histories of hypotonia through childhood that impeded motor development and even feeding ability in early infancy, and sensory integration problems, but normal cognitive abilities. Neither ACC nor other ectodermal changes are noted, but the youngest brother (II-3) has unilateral cleft hand and polydactyly. More subtle foot, toe, and other minor digital anomalies vary among the three affected males. All three also have histories of cryptorchidism and/or hypospadias. Each is heterozygous for a de novo frameshift UBA2 variant, c.1376_1377insT, p.Thr460Aspfs*24, not detected in blood of either parent with either next-generation (130× coverage at 10× depth) or Sanger sequencing.

Family 3: Clinical details about part of this family were reported previously14 but are now updated and expanded along with results of exome analysis. The male proband (II-2, Fig. 1a, b) has a single area of ACC, supernumerary nipple, cryptorchidism, early developmental delay, astigmatism, learning disability, depression, bipolar disorder, and social phobia. His mother (I-2) has multiple areas of healed ACC, supernumerary nipples, small head circumference, and asymmetric kidneys with reduced renal function. Neither have documented hand or foot anomalies. They are both heterozygous for a nonsense variant in UBA2: c.364C>T, p.Arg122*. Two other affected individuals (II-1 and III-1) have similar facial features, ACC, and supernumerary nipples and were each confirmed to harbor the familial UBA2 variant.

Family 4: The female proband (II-1, Fig. 1b), 21 years old at examination, has a history of delayed motor skills and attention deficit disorder. Height, weight, and head circumference are all currently less than the third percentile; she also had early growth deficiency, delayed dentition and bone age. Features include ACC, thin scalp hair, clinodactyly, and overlapping toes. See Table 1 and supplement for additional endocrine, renal, and ophthalmologic concerns. She is heterozygous for a de novo missense variant in UBA2, c.167A>C: p.Asn56Thr.

Family 5: The female proband (Fig. 1b, II-1), 4 years 9 months old at exam, has developmental delay, absent speech, hemangiomas, ACC, and seizures. She has relative macrocephaly, epicanthal folds, anteriorly placed anus, and pes planus. She carries a de novo missense UBA2 variant: c.1447G>A, p.Glu483Lys.

Family 6: The proband is a male toddler (Fig. 1a, b, II-1) with cryptorchidism, bilateral inguinal hernias, and multiple limb deformities including bilateral ectrodactyly of the feet, complete 2–3 finger syndactyly, clinodactyly, and camptodactyly. He has low-normal growth and normal developmental milestones. Facial features include hypertelorism, bilateral epicanthal folds, and pseudostrabismus. He does not have ACC or other ectodermal abnormalities. He is heterozygous for a de novo UBA2 nonsense variant: c.800T>A, p.Leu267*.

Family 7: The proband (Fig. 1b, II-1) is a 3-year, 11-month-old Caribbean male born at 35 weeks gestational age. At two weeks, height and weight (corrected for prematurity) were normal, but head circumference measured at the 2nd centile. He had global developmental delay and four limb ectrodactyly, tall and prominent forehead, deep-set eyes, broad nasal root, left preauricular tag, narrow palate, and a vertical cleft chin. Presurgery, he had left 2–3 finger syndactyly with a nodule adjacent to the medial aspect of the PIP joint of the 4th finger. The right 3rd digit is missing; other digits are relatively normal. On the left foot, two malformed digits are divided by a deep central cleft; the right foot also has a deep central cleft with three malformed digits, and 4–5 toe syndactyly. He does not have ACC but has large areas of faint hypopigmentation over his torso and limbs. He is heterozygous for a de novo missense variant in UBA2: c.364C>G, p.Arg122Gly.

None of the detected UBA2 variants was found in the gnomAD database.15 Results of in silico predictor analyses for missense variants and variant classification are provided in Supplemental Tables 1 and 2. All would be classified as pathogenic or likely pathogenic using American College of Medical Genetics and Genomics (ACMG)/Association for Molecular Pathology (AMP) guidelines (classification criteria)16 in Supplemental Table 2.

Modeling effects of missense variants on UBA2 function

UBA2 in complex with SAE1 plays a key role in the SUMOylation pathway. Observed human UBA2 variants are distributed across the gene (Fig. 2a, b). All truncating variants are expected to undergo nonsense-mediated decay based on their position within the mRNA. Missense variants occur at residues that are strongly conserved across vertebrates (Fig. 2c). Given the similarities in phenotypes between individuals with truncating and missense alleles, we hypothesized that missense alleles also lead to loss of function.

To understand how missense alleles might disrupt UBA2 function, molecular modeling using published crystal structures17 and simulated substitutions were performed for each detected human missense variant. In the UBA2 protein, p.Gly2418 is directly involved in ATP binding; its substitution with valine results in altered protein conformation and is predicted to result in loss of ATP binding and ectopic interactions with nearby residues (Fig. 2d).17 Similarly, asparagine replacement with threonine at position 56 putatively abolishes ATP-dependent activation. The p.Arg122Gly substitution is predicted to result in loss of interaction with ATP. Human UBA2 protein interacts with a conjugating enzyme called UBC9 (amino acids 6–38) via amino acid residues 478–509, which include Glutamate 483. UBA2 forms a hydrophobic bond with Leu6, Met36, and Leu38 of UBC9; replacing Glutamate 483 with Lysine is predicted to disrupt UBA2–UBC9 binding. In summary, missense alleles observed in patients with UBA2-associated syndrome are observed to occur at functionally critical residues and potentially disrupt ATP binding, protein folding, or protein–protein interactions.

Zebrafish uba2 expression in affected tissues

By whole-mount in situ hybridization, uba2 transcript was detected on the dorsoventral axis of 5-somite stage embryos (Fig. S1a, b). At later stages, uba2 is expressed in developing brain, eye, craniofacial structures, and fins. At 24 hpf, uba2 expression was restricted to the head region, including the eye and nervous system (Fig. S1c). At 35 hpf, prominent signal was observed in pectoral fins (arrows, Fig. S1d). At all other examined stages (48 and 72 hpf, 5 and 7 dpf), uba2 mRNA signal localized to the head region, specifically brain, neural retina, and lens (Fig. S1e–h). Therefore, zebrafish uba2 is expressed in some structures that are analogous to those affected in humans harboring deleterious UBA2 variants.

Variable expressivity observed with uba2 loss of function

uba2 knockout zebrafish lines were generated by CRISPR/Cas9-targeted deletion. The phenotype of homozygous fish was notable for failure to inflate swim bladders. At 5–8 dpf, we observed severe gross morphological defects in uba2-/- zebrafish (Fig. 3) including small eyes, hydrocephalus and craniofacial edema, ventrally curved body axis, and uninflated swim bladder. Faint heartbeat and severe pericardial edema were observed in 41% of embryos (Fig. 3a, b). Edema became generalized at 8 dpf when most lethality was noted. To further examine the effect of uba2 on zebrafish development, we calculated the survival rate of uba2-/- zebrafish which was significantly lower than control (WT) and heterozygous fish. uba2-/- zebrafish showed a mortality rate of approximately 50% at 8 dpf; however, 100% of mutant fish were dead by day 12 (Fig. 3d).

Fig. 3: Severe dysmorphic features in embryonic uba2-/- zebrafish.
figure 3

(a, b) Dorsal and lateral views of uba2 zebrafish at 8 dpf. Compared to wild-type (WT) controls, mutant lines showed aberrant head development with small eyes and hydrocephalus. (c) Bar graph representing the percentage of uba2-/- zebrafish with gross morphological defects. (d) Survival curve showing the number of live fish over the course of 21 days. WT and heterozygous fish showed similar death curves, but homozygous fish had steeper death curves with 100% mortality by day 12. B brain, F craniofacial, PF pectoral fins, SB swim bladder, T tail.

Full size image

Nullizygous fish exhibited a wide phenotypic range. We observed a pair of normal extended pectoral fins in WT zebrafish versus uba2-/- fish, where pectoral fins were found to be short and upright-oriented (Fig. 3a) confirming uba2 function in fish extremity development. WT zebrafish had thin lines originating from base to fin tips showing normal actinotrichia. In contrast, uba2-/- fish displayed collapsed (Fig. 3b, middle image) and irregular fin fold edges (Fig. 3b, last image).

To better characterize variable expression and the relationship between the zebrafish knockout and the human disorders, we quantified craniofacial (F), brain (B), pectoral fin (PF), tail fin (TF), and swim bladder (SB) defects. Defects at later stages of development were studied in uba2-/- fish bred from the same parent at 8 dpf, when approximately half the fish survive (n = 32; Fig. 3c). Tissue-level malformations were observed in craniofacial structures (9.38%), brain size (90.6%), tail fin (25%), pectoral fin (100%), and swim bladder (93.75%) (Fig. 3c and as described below). Thus, across individual fish with similar genetic backgrounds, total uba2 function loss recapitulates some tissue-level phenotypes and the variable expression observed in human UBA2-related phenotypes.

Neuronal reduction in uba2 zebrafish

Tissue-level analysis was performed in zebrafish to elucidate abnormalities resulting from uba2 loss of function. First, we conducted immunohistochemistry studies on 8 dpf zebrafish cryosections through eye and brain. Compared to WT controls, uba2-null fish showed small heads, reduced midbrain size, low nuclei cell count with high accumulation of actin signal (orange, Fig. S2), implying a decreased proportion of gray to white matter. In addition, uba2-/- fish had smaller eyes, reduced retinal thickness, retinal laminations, and lens defects (see Supplement).

Skeletal and extremity phenotypes in the uba2 zebrafish model

To investigate the impact of uba2 on zebrafish skeletal development, we stained uba2 WT (+/+), heterozygous (+/-), and homozygous (-/-) fish with alcian blue dye at 5 dpf. In both uba2 WT (Fig. 4a) and heterozygous zebrafish (data not shown), alcian staining demonstrated a normal pattern of cartilage element development including typical ceratohyoid, Meckel’s cartilage, ceratobranchials arches, and pectoral fin cartilage. However, complete loss of uba2 in homozygous fish resulted in abnormal craniofacial development. In addition to jaw malformations, other craniofacial malformations included malformed and hypoplastic ventral and dorsal cartilage structures with lack of basihyal and hypohyal development. We also noted an apparently abnormal fusion of Meckel’s cartilage with the palatoquadrate, resulting in a small, narrow mandible (Fig. 4b). Moreover, Meckel’s cartilage was flattened at the midline fusion point with completely absent ceratohyal cartilage and ceratobranchials arches, the equivalent of micrognathia in these fish.

Fig. 4: Cranial cartilage patterns observed in uba2-/- zebrafish and rescue of uba2 mutant phenotype with human UBA2 messenger RNA (mRNA).
figure 4

(a, b) Brightfield ventral and lateral views of cartilage stained uba2 in wild type (WT) and homozygous mutant fish are shown in top and bottom panels, respectively. Closeups of pectoral fin cartilage phenotype are shown in inserts in the bottom panel highlighted by black dashed boxes on lateral views. an anterior, cb ceratobranchials 1–4, ch ceratohyal, ep ethmoid plate, h hypohyal, hs hyosymplectic, mk Meckel’s cartilage, pq palatoquadrate. (c) Z-stack images of uba2 zebrafish median fins stained with Col2a (green), Rhodamine-Phalloidin (red), and Dapi (blue). Arrows are used to show the gaps between actinotrichia fibers. Scale bar: 50 µm. (d) Suppression of uba2 in zebrafish produces an abnormal phenotype that is classified into three categories. (e) Proportions of uba2-/- zebrafish embryos representing each phenotype category after injecting with WT or mutation harboring human UBA2 mRNA. Landmark abbreviations: Ab abnormal, Nor St normal structure, ns not significant. Chi Square test p values are shown above the phenotypes for each rescue experiment.

Full size image

To explore whether uba2 mutation causes skeletal phenotypes in adult fish, we performed microcomputed tomography (CT) comparing WT (n = 3) and uba2+/- (n = 3) fish, as nullizygous fish did not survive to this stage. We noted abnormal, wavy ribs and dysmorphic fin girdles in uba2+/- fish (Fig. S3).

In teleosts, finfolds are typically made of type II collagen matrix structures called actinotrichia that line the epidermis. Brightfield microscopy of uba2-/- fish revealed structural defects in median fins (Fig. 3b). To examine the effect of uba2 truncation on zebrafish median fin structure development, we stained uba2 zebrafish (+/+, +/-, and -/-) larvae with type II collagen (Col2a) and Phalloidin (F-actin) antibodies to label actinotrichia (Fig. 4c).

Actinotrichia fibrils initiate fin development and become the future fin connective tissue. At 5 dpf, both WT (Fig. 4c, top panel) and heterozygous (data not shown) larvae develop median fins showed normally arrayed Col2a-labeled actinotrichia fibers; however, we observed nonrigid, nonparallel and bent actinotrichia in uba2-/- (Fig. 4c, arrows) fish. Phalloidin staining in uba2-/- fish revealed disorganized and disrupted organization, corresponding to areas of this abnormal collagen pattern (Fig. 4c).

Further investigating these extremity defects at a cellular level, we performed ultrastructural analysis of the uba2 zebrafish body wall near the median finfold at 5 dpf. Detailed examination by TEM revealed a typical dynamically assembled dense striated pattern of actinotrichia in WT fish (Fig. S4). Similarly, in WT fish we observed a normal and organized distribution of skeletal muscles with normal nuclei and mitochondria. However, in uba2-/- zebrafish, we observed disorganized (or incompletely developed) and scattered actinotrichia with abnormal epidermal cells (arrow). The skeletal muscle layer in homozygous fish was also observed to be discontinuous or atrophic with degenerated nuclei and mitochondria. Therefore, absent uba2 impacts connective and epithelial tissue and skeletal muscle and causes extremity malformations in developing fish.

Conserved function of UBA2 candidate variants in zebrafish

To further confirm the specificity of the uba2 knockout phenotype, we attempted phenotypic rescue of developmental fish malformations by injecting human UBA2 mRNA. Injected fish were grouped into three phenotypic classes and genotyped at 5 dpf, and the uba2-/- subset was analyzed. Embryos were classified as class I (grossly normal body structure), class II (decreased head size, absent swim bladder), and class III (small head and body, generalized edema) (Fig. 4d). As compared to H2O-injected controls, injecting human WT UBA2 mRNA grossly rescued phenotypes in a significant number of fish. The proportion of class I fish increased from 5% to 33%, and the proportion of class III fish decreased from 47% to 6% (p < 0.0001) (Fig. 4e). Even though WT UBA2 mRNA injection rescued gross phenotypes, most uba2-/- zebrafish still did not show inflated swim bladder (data not shown), suggesting that early uba2 deficiency permanently impacts zebrafish physiology despite substitution with human mRNA.

Human mRNAs encoding p.Gly24Val, p.Arg122Gly and p.Glu483Lys all failed to rescue the uba2-/- phenotypes in contrast to WT mRNA. The p.Asn56Thr substitution demonstrated statistically similar rescue to control mRNA; however, there were more class III fish (23% vs. 6%) and fewer class I fish (18% vs. 33%) following p.Asn56Thr injection, indicating possible partial loss of function for this missense substitution (Fig. 4e). Because the mRNAs containing the missense variants failed to rescue uba2-null phenotypes to a similar level as did WT UBA2 mRNA, we conclude that the most likely mechanism of disease is loss of function.

DISCUSSION

In this study, we describe a cohort of patients harboring deleterious variants in the UBA2 gene. They show highly variable inter- and intrafamilial expression of dermatologic, skeletal, extremity, neurologic, cardiac, and renal features, similar to those of the chromosome 19q13.11 microdeletion syndrome.1,2,3,4,5,6,7,8 These observations further support UBA2 as the critical gene in the microdeletion syndrome and suggest its essential role in early human growth and development. There are only a few other reports of intragenic UBA2 variants (summarized in Table 1). Marble et al.18 reported a de novo UBA2 missense variant (c.71G>T, p.Gly24Val) in a 2.5-year-old female with ACC, thin hair, tall forehead, Duane anomaly, hip dysplasia, clinodactyly, and poor weight gain. Wang et al.18 reported an inherited UBA2 frameshift variant (c.327delT, p.Phe109Leufs*3) in a young boy and his mother. The mother had ACC but was otherwise healthy. The son had ACC, microcephaly, bilateral ectrodactyly, low‐lying conus medullaris, horseshoe kidney, and tracheoesophageal fistula. A de novo UBA2 loss-of-function variant (c.1324dupT, p.Tyr442Leufs*17) was associated with four extremity split hand and foot malformation with tibial deficiency and undermasculinized external genitalia.19 Aerden et al.20 reported a male proband with ectrodactyly of the feet, autism spectrum disorder, craniofacial variations, dry sparse scalp hair, strabismus, and hypermetropia who was heterozygous for a de novo frameshift variant in UBA2 (c.612delA, p.Glu205Lysfs*63); this was considered to be responsible for the phenotype.20

The four patients previously reported with intragenic UBA2 variants were added to our clinical summary table (Table 1) to compare phenotypes.18,19,20,21 We’ve estimated the percentage of key traits in UBA2 subjects (Fig. 1c) based on available clinical information. The most specific aspects of the UBA2-related phenotype are ACC, seen in 61%, and ectrodactyly, which is less common (37%). Early growth deficiency and neurodevelopmental delay are reported in 61% and 80% of affected individuals, respectively. More variable digital and skeletal abnormalities are also present (56%) but are sometimes subtle and potentially overlooked (e.g., Fig. 1a, panels C, D). These include clinodactyly (62%), syndactyly (59%), camptodactyly (57%), and hip abnormality (35%). The most common craniofacial variations are tall forehead/high hairline (76%), downslanted palpebral fissures (47%), hypertelorism (62%), broad nasal root (81%), microcephaly (37%), and micrognathia (53%). Other observed features among our subjects include other ectodermal variations (~82%), ocular abnormalities (53%), and cardiac (43%), genital (50%, in males), and renal (36%) abnormalities.

In C. elegans, Uba-2 is also noted to be a critical element of the SUMOylation pathway; its ablation leads to embryonic lethality.22 UBA2 acute knockdown in xenograft tumors by conditional short hairpin RNA (shRNAs) causes marked growth arrest, cell proliferation defects, and increased apoptosis.23 In mice, loss of any key component of the SUMOylation pathway can lead to severe impairment of cellular functions and lethality.24,25 An in situ hybridization study conducted in mouse embryos (8.5 to 11.5 days postcoitum) revealed Uba2 ample expression at multiple morphogenetic activity sites, e.g., neural folds, branchial arches, and limb buds,24 suggesting that Uba2 is essential for normal cellular function/development. Recently, SUMOylation was reported to regulate differentiation of several ocular tissues.26,27

Phenotypic features in our human UBA2-related syndrome cohort and the uba2 knockout zebrafish are reminiscent of disorders associated with pathogenic variants in DLX5/6 (split hand/foot malformation [SHFM1], OMIM 220600), TP63 (e.g., ectrodactyly, ectodermal dysplasia, and cleft lip/palate syndrome 3 [EEC3], OMIM 604292; split hand/foot malformation syndrome 4 [SHFM4], OMIM 605289; and others), and FBXW4, a candidate for SHFM3 (OMIM 246560). tp63-/- zebrafish embryos have ectodermal defects involving skin, absent pectoral fin buds, and reduced size fin folds at 36 hpf and embryos died between 40–50 hpf.28 tp63 zebrafish morphants affect skin integrity by making the skin more prone to microbial infection.29 fndc3a-/- zebrafish show broken actinotrichia, aberrant collagen localization, and cellular defects in epidermal cells during caudal fin development.30 It is possible that these genes function downstream of the SUMOylation pathway, leading to phenotypes that overlap the UBA2-related syndrome.

In the current study, the mRNA rescue experiments showed that WT UBA2 mRNA injection partially rescued the abnormal head/eye, tail, and uninflated swim bladder phenotype in uba2-/- zebrafish (33%). Notably, three of four human missense UBA2 mRNAs did not rescue the uba2-/- phenotype to a significant degree, suggesting a loss-of-function mechanism for these disease-associated alleles. As wide phenotypic variability is observed in both fish and human UBA2/uba2-related phenotypes, additional studies are warranted to define potential modifiers. Morpholinos (MOs) have been used in reverse genetic studies in a range of animal models.31,32 However, MOs may be hard to interpret as they typically result in more severe phenotypes.33 mRNA rescue in CRISPR-generated stable mutant lines is potentially useful in the interpretation of MO-related inconsistencies. Precise single-nucleotide variant animal models of human diseases can help to better understand underlying molecular processes and may aid in management of UBA2-related abnormalities.34

In conclusion, we report clinical details in 16 individuals from seven unrelated families with inherited or de novo heterozygous UBA2 sequence variants, who present with highly variable phenotypes. Definition of the UBA2-related autosomal dominant phenotypic spectrum in humans, in silico modeling predictions, uba2 expression, and characterization of the knockout phenotype in zebrafish support the significance of UBA2/uba2 in development, potentially by affecting post-translational modification of SHFM-associated genes. mRNA rescue experiments in zebrafish also suggest that loss of gene function is the primary mechanism of disease. The highly variable expressivity of the human UBA2 phenotype, either via sequence alteration or contiguous gene deletion, even within the same family, remains incompletely explained; there are likely other modifiers, still to be identified. However, our studies define a human disorder associated with UBA2 sequence variants with a phenotype that overlaps key aspects of the chromosome 19q13.11 microdeletion syndrome.

Web Resources

ClinVar Database https://www.clinicalgenome.org/data-sharing/clinvar

gnomAD https://gnomad.broadinstitute.org/

GeneMatcher https://genematcher.org/

Pathogenicity predictions https://varsome.com/

OMIM http://www.omim.org/

Clustal omega https://www.ebi.ac.uk/Tools/msa/clustalo/