*Article* **The Clinical and Genotypic Spectrum of Scoliosis in Multiple Pterygium Syndrome: A Case Series on 12 Children**

**Noémi Dahan-Oliel 1,2,\*, Klaus Dieterich <sup>3</sup> , Frank Rauch 1,2, Ghalib Bardai 1,2, Taylor N. Blondell <sup>4</sup> , Anxhela Gjyshi Gustafson <sup>5</sup> , Reggie Hamdy 1,2, Xenia Latypova <sup>3</sup> , Kamran Shazand <sup>5</sup> , Philip F. Giampietro <sup>6</sup> and Harold van Bosse 4,\***


**Abstract:** Background: Multiple pterygium syndrome (MPS) is a genetically heterogeneous rare form of arthrogryposis multiplex congenita characterized by joint contractures and webbing or pterygia, as well as distinctive facial features related to diminished fetal movement. It is divided into prenatally lethal (LMPS, MIM253290) and nonlethal (Escobar variant MPS, MIM 265000) types. Developmental spine deformities are common, may present early and progress rapidly, requiring regular fo llow-up and orthopedic management. Methods: Retrospective chart review and prospective data collection were conducted at three hospital centers. Molecular diagnosis was confirmed with whole exome or whole genome sequencing. Results: This case series describes the clinical features and scoliosis treatment on 12 patients from 11 unrelated families. A molecular diagnosis was confirmed in seven; two with *MYH3* variants and five with *CHRNG*. Scoliosis was present in all but our youngest patient. The remaining 11 patients spanned the spectrum between mild (curve ≤ 25◦ ) and malignant scoliosis (≥50◦ curve before 4 years of age); the two patients with *MYH3* mutations presented with malignant scoliosis. Bracing and serial spine casting appear to be beneficial for a few years; non-fusion spinal instrumentation may be needed to modulate more severe curves during growth and spontaneous spine fusions may occur in those cases. Conclusions: Molecular diagnosis and careful monitoring of the spine is needed in children with MPS.

**Keywords:** *CHRNG*; distal arthrogryposis type 8; Escobar; multiple pterygium syndrome; *MYH3*; scoliosis

### **1. Introduction**

The term pterygium is used most commonly to describe an acquired ophthalmologic condition, with a conjunctival "wing" or flap that can cross the cornea. The term also describes joints with congenital webbing or winging of soft tissue which limits the joint motion. Usually, more than one joint and body region is involved, meeting the criteria for arthrogryposis multiplex congenita [1]. While pterygia may be an incidental finding in persons with arthrogryposis, such as the occasional knee pterygium found associated with severe contractures in Amyoplasia, they can also manifest as a more generalized syndrome, such as popliteal pterygium syndrome or multiple pterygium syndrome (MPS).

MPS is a rare form of arthrogryposis multiplex characterized by a constellation of congenital anomalies [2]. The webbing of skin and contractures of the joints that are found in this disorder may restrict movement. Examples of joint involvement in the

**Citation:** Dahan-Oliel, N.; Dieterich, K.; Rauch, F.; Bardai, G.; Blondell, T.N.; Gustafson, A.G.; Hamdy, R.; Latypova, X.; Shazand, K.; Giampietro, P.F.; et al. The Clinical and Genotypic Spectrum of Scoliosis in Multiple Pterygium Syndrome: A Case Series on 12 Children. *Genes* **2021**, *12*, 1220. https://doi.org/ 10.3390/genes12081220

Academic Editor: Stephen Robertson

Received: 18 June 2021 Accepted: 2 August 2021 Published: 6 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

upper extremities include axillary pterygia (both the anterior and posterior folds), elbow flexion contractures with antecubital webbing, mildly dorsiflexed wrists, and fingers with camptodactyly, interdigital pterygia and thumb-in-palm deformities. In the lower extremities, perineal pterygia can span medially from one thigh to the other, knee flexion contractures with pterygia can be severe, and foot deformities include both clubfoot and congenital vertical talus (rocker bottom foot). Other characteristic findings of MPS are short stature, webbing of the neck, and distinctive facial features including micrognathia, cleft palate, down-turned corners of the mouth, an elongated philtrum, down-slanting palpebral fissures, epicanthal folds, ptosis, and low-set ears, all of which are related to diminished fetal movement. Developmental spine deformities are common, not only as a coronal plane deformity (scoliosis), but fre-

MPS is a rare form of arthrogryposis multiplex characterized by a constellation of congenital anomalies [2]. The webbing of skin and contractures of the joints that are found in this disorder may restrict movement. Examples of joint involvement in the upper extremities include axillary pterygia (both the anterior and posterior folds), elbow flexion contractures with antecubital webbing, mildly dorsiflexed wrists, and fingers with camptodactyly, interdigital pterygia and thumb-in-palm deformities. In the lower extremities, perineal pterygia can span medially from one thigh to the other, knee flexion contractures with pterygia can be severe, and foot deformities include both clubfoot and congenital

*Genes* **2021**, *12*, 1220 2 of 14

vertical talus (rocker bottom foot).

Other characteristic findings of MPS are short stature, webbing of the neck, and distinctive facial features including micrognathia, cleft palate, down-turned corners of the mouth, an elongated philtrum, down-slanting palpebral fissures, epicanthal folds, ptosis, and low-set ears, all of which are related to diminished fetal movement. Developmental spine deformities are common, not only as a coronal plane deformity (scoliosis), but frequently with a substantial associated sagittal plane deformity, making it a kyphoscoliosis. Spontaneous spinal fusion abnormalities occur often in MPS and are congenital. quently with a substantial associated sagittal plane deformity, making it a kyphoscoliosis. Spontaneous spinal fusion abnormalities occur often in MPS and are congenital. MPS can be separated into the lethal pterygium syndromes (LMPS, MIM253290) and the non-lethal syndromes; the latter conditions are categorically referred to as Escobar syndrome (or Escobar variant MPS, MIM265000) and most are autosomal recessive. Here we will only be discussing the non-lethal forms of MPS.

MPS can be separated into the lethal pterygium syndromes (LMPS, MIM253290) and the non-lethal syndromes; the latter conditions are categorically referred to as Escobar syndrome (or Escobar variant MPS, MIM265000) and most are autosomal recessive. Here we will only be discussing the non-lethal forms of MPS. Several genes associated with MPS give rise to what has recently been described as a prenatal form of myasthenia, first associated with variants in *CHRNG*. *CHRNG* codes for the γ subunit of the acetylcholine receptor (AChR) in the developing fetus. Mutations that

Several genes associated with MPS give rise to what has recently been described as a prenatal form of myasthenia, first associated with variants in *CHRNG*. *CHRNG* codes for the γ subunit of the acetylcholine receptor (AChR) in the developing fetus. Mutations that impact the expression of this subunit, its integration in the AChR, or the transport of the AChR to the sarcolemma will have major consequences to the neuromuscular junction. At 33 weeks gestation though, the γ subunit starts to be replaced by the *e* subunit of AChR, finally leading to a functional adult neuromuscular junction. However, effects of the fetal akinesia have become manifest before that stage of development and the damage is irreversible. Interestingly, individuals with Escobar syndrome do not have muscle weakness or electrophysiological symptoms associated with myasthenia gravis postnatally, as the AChR essentially functions normally after birth. Other causes of Escobar syndrome recessive gene mutations of the AChR include *CHRNA1* (α1-AChR subunit), *CHRNB1* (β1-subunit), *CHRND* (δ-subunit), and *RAPSN* (AChR binding protein). Escobar syndrome can also be caused by pathogenic variants in *CNTN1* (contactin 1) and *DOCK7* (dedicator of cytokinesis 7) [3–5]. See Figure 1. impact the expression of this subunit, its integration in the AChR, or the transport of the AChR to the sarcolemma will have major consequences to the neuromuscular junction. At 33 weeks gestation though, the γ subunit starts to be replaced by the ɛ subunit of AChR, finally leading to a functional adult neuromuscular junction. However, effects of the fetal akinesia have become manifest before that stage of development and the damage is irreversible. Interestingly, individuals with Escobar syndrome do not have muscle weakness or electrophysiological symptoms associated with myasthenia gravis postnatally, as the AChR essentially functions normally after birth. Other causes of Escobar syndrome recessive gene mutations of the AChR include *CHRNA1* (α1-AChR subunit), *CHRNB1* (β1-subunit), *CHRND* (δ-subunit), and *RAPSN* (AChR binding protein). Escobar syndrome can also be caused by pathogenic variants in *CNTN1* (contactin 1) and *DOCK7* (dedicator of cytokinesis 7) [3–5]. See Figure 1.

**Figure 1.** Clinical spectrum and overlap of molecular causes of arthrogryposis multiplex congenita (AMC). The main categories of AMC phenotypes are represented as dark grey boxes. Some of the genetic determinants that are responsible for multiple phenotypes among AMC subtypes are represented in white boxes. Both *MYH3* and *TPM2* biallelic loss of function have been associated to MPS, although dominant heterozygous variants of these genes are linked to several forms of distal arthrogryposis. FADS: fetal akinesia deformation sequence; MPS: multiple pterygium syndrome. **Figure 1.** Clinical spectrum and overlap of molecular causes of arthrogryposis multiplex congenita (AMC). The main categories of AMC phenotypes are represented as dark grey boxes. Some of the genetic determinants that are responsible for multiple phenotypes among AMC subtypes are represented in white boxes. Both *MYH3* and *TPM2* biallelic loss of function have been associated to MPS, although dominant heterozygous variants of these genes are linked to several forms of distal arthrogryposis. FADS: fetal akinesia deformation sequence; MPS: multiple pterygium syndrome.

More recently, cases of MPS with an autosomal dominant form, also classified as distal arthrogryposis type 8 (DA8) [6,7], have been associated with mutations in the embryonic myosin heavy chain gene, *MYH3*. This is the same gene that is found underlying other forms of distal arthrogryposis (DA1, DA2A or Freeman Sheldon syndrome, and DA2B or Sheldon Hall syndrome).

Spinal curvatures in Escobar syndrome appear early, often present at birth, and can progress quickly. Treatment options include spine casting, bracing, and expandable implant surgery, to allow as much chest growth and development as possible, and ultimately spinal fusion, with a goal of well-balanced spine [8]. Although advances in the phenotypic spectrum, disease progression and genetic etiology of MPS have been made [9,10], definite phenotype–genotype correlations need yet to be discovered. The objective of this case series was to describe the phenotypic presentations of a small multisite collection of patients with Escobar syndrome, in particular as to how it relates to their spinal deformity. By determining the genetic basis of their MPS, we hope to associate the genotype with their phenotype and natural history of their scoliosis, and describe the clinical interventions aimed at reducing the spinal curvatures associated with these cases.

### **2. Materials and Methods**

### *2.1. Design*

Case series using a retrospective chart review and prospective data collection at three hospital centers; Shriners Hospitals for Children in Montreal, Canada, Shriners Hospitals for Children in Philadelphia, United States, and Grenoble Alpes University Hospital Center in France. The patients at the two Shriners hospitals were enrolled in a multisite pediatric arthrogryposis registry that collects clinical, patient-reported outcomes, and provides whole genome sequencing to map the phenotype to genotype in this population.

### *2.2. Ethical Approval and Consent*

Written informed consent according to local ethical committees in all participating centers was obtained for all patients. Clinical and genetic data were anonymized and entered into a secured database accessible only to the research team.

### *2.3. Patients*

We describe the phenotypic presentation of 12 patients presenting with MPS and scoliosis. Clinical data were analyzed retrospectively. Clinical data (sex, age), scoliosis (i.e., age at onset, congenital/early onset/late onset), and a description of the phenotype was extracted from the electronic medical record and evaluated in clinic (contractures, limb anomalies, webbing/pterygia, other system involvement). A description of interventions, including type (observation, bracing, and surgery) and age at treatment was collected. Spine X-rays were reviewed and classified in order to better understand the curves in children with Escobar syndrome using the following classification system of curve behavior. A mild curve was defined as a curve ≤ 25◦ at any age. A moderate curve was a curve > 25◦ but ≤50◦ at any age. Severe curves exceeded 50◦ , but not until after 4 years of age, whereas malignant curves were ≥50◦ before 4 years of age.

### *2.4. Genotyping and Data Analysis*

Genomic DNA was isolated from saliva collected in Oragene OGR-600 tubes (DNA Genotek, Kanata, ON, Canada) according to manufacturer's instructions and subsequently extracted using the Chemagic360 platform (PerkinElmer, Waltham, MA, USA). DNA was quantified using Qubit. Average gDNA obtained from extractions was 40 ng/uL. Libraries were prepared in the Zephyr G3 NGS Automated Workstation (PerkinElmer) using the Illumina DNA PCR-Free Prep, Tagmentation (Illumina, San Diego, CA, USA) with 1000 ng of input gDNA. Resulting libraries were quantified prior to pooling using the Kapa Library Quantification Kit (Roche, Basel, Switzerland). Libraries were pooled at a concentration of 1.4 nM and were sequenced on the NovaSeq 6000 sequencing platform (Illumina). Paired-

end sequencing (2 × 150 bp) was performed at >30× coverage per sample. GRCh38 reference genome was used. Genotyping at the Grenoble arthrogryposis reference center was conducted using DNA extracted from whole blood samples collected on EDTA tubes. Libraries were prepared with the Nextera DNA Flex Library Prep Kit (Illumina). Exome sequencing was performed on a NextSeq 500 (Illumina).

Data analysis, including variant calling, was performed using the Illumina TruSight™ Software Suite platform. This platform performs alignment, variant calling, variant annotation and filtering. For the Grenoble hospital center, bioinformatic analysis was performed with an in-house pipeline using an NGS platform, using the Burrows-Wheeler Aligner, Picard Tools 2.18.23-0 and GATK v4.0.12. Variants were interpreted with ANNOVAR and prioritized through python annotation scripts. The strategy for genome data interpretation was primarily based on disease and phenotype gene target definition. Data were descriptively analyzed.

### **3. Results**

### *3.1. Clinical Features*

The 12 patients (7 females) with MPS and scoliosis were from 11 unrelated nonconsanguineous families. They were born in Algeria (*n* = 1), Canada (*n* = 1), France (*n* = 4) and the United States (*n* = 6) and were between 1 and 20 years old at the time of last follow up (Tables 1 and 2). Family history was positive in one family (F6), with presence of clinical features of Escobar and scoliosis in the mother and maternal aunt of the index patient (P7). In the family of two affected siblings (F2), one pregnancy had been terminated due to severe fetal deformities; muscle histology in that fetus was in accordance with a diagnosis of Escobar syndrome.


### **Table 1.** Phenotype: pterygia and joints affected.

The children were born after 34–41 weeks of pregnancy (Table 3). Six of the children were born vaginally, while the other six were born via c-section due to breech position and complicated pregnancy with oligohydramnios or polyhydramnios. Eight children were admitted to the neonatal intensive care unit after birth due to feeding difficulties and/or respiratory issues.

Characteristic phenotypic features of Escobar were observed in all 12 patients, including webbing over the neck, axillary, elbow, knee and/or fingers (Table 1). Downslanting palpebral fissures were observed in 10 patients (Table 2). Three patients had cleft palate and another five had high palate. Four patients had both posterior rotated ears

and low-set ears, whereas another three patients either had posterior rotated ears or lowset ears. Figures 2 and 3 showcase pterygia and characteristic features in MPS among two patients.


**Table 2.** Phenotype: facial features and other characteristics.

Seven of the 12 patients had foot deformities, six with congenital vertical tali and one with clubfeet (Table 1). Contractures were widespread across several joints (e.g., shoulders, elbows, wrists, fingers, hips, knees, ankles) and varied in severity among patients. Regarding functional mobility (data available in 11 patients), five patients were independent walkers, one used a stroller as she did not crawl or walk yet at age 22 months, two used a manual wheelchair for outdoor mobility, and three used a motorized chair for outdoor mobility. Of the five children who ambulated without a mobility aid, four wore knee ankle foot orthoses and two walked with a crouched gait.


**Table 3.** Pre- and postnatal information.

IUGR: intra uterine growth restriction; NICU: neonatal intensive care unit. patients.

**Figure 2.** Patient 1 with micrognathia, downslanting palpebral fissures, lowset and posteriorly rotated ears, and pterygia apparent in the antecubital and knee popliteal regions. **Figure 2.** Patient 1 with micrognathia, downslanting palpebral fissures, lowset and posteriorly rotated ears, and pterygia apparent in the antecubital and knee popliteal regions.

**Figure 3.** Patient 10 with mild to moderate axillary, popliteal, and inguinal pterygia.

ankle foot orthoses and two walked with a crouched gait.

Seven of the 12 patients had foot deformities, six with congenital vertical tali and one with clubfeet (Table 1). Contractures were widespread across several joints (e.g., shoulders, elbows, wrists, fingers, hips, knees, ankles) and varied in severity among patients. Regarding functional mobility (data available in 11 patients), five patients were independent walkers, one used a stroller as she did not crawl or walk yet at age 22 months, two used a manual wheelchair for outdoor mobility, and three used a motorized chair for outdoor mobility. Of the five children who ambulated without a mobility aid, four wore knee

Scoliosis was present in all but one patient who was only nine months of age at the last assessment. Four patients had malignant curves that had developed either during infancy or possibly prenatally (prenatal scoliosis). The level of the major curve was at the thoracolumbar spine in two of these patients, the other two patients had left low thoracic curves. These curves all included the pelvis, causing moderate to severe pelvic obliquity (18–50°). One patient had a mild but sweeping kyphosis including the pelvis, two had apical mid thoracic hyperkyphosis, whereas the last had thoracic hypokyphosis but with patients.

apparent in the antecubital and knee popliteal regions.

**Figure 2.** Patient 1 with micrognathia, downslanting palpebral fissures, lowset and posteriorly rotated ears, and pterygia

**Figure 3.** Patient 10 with mild to moderate axillary, popliteal, and inguinal pterygia. **Figure 3.** Patient 10 with mild to moderate axillary, popliteal, and inguinal pterygia.

Seven of the 12 patients had foot deformities, six with congenital vertical tali and one with clubfeet (Table 1). Contractures were widespread across several joints (e.g., shoulders, elbows, wrists, fingers, hips, knees, ankles) and varied in severity among patients. Regarding functional mobility (data available in 11 patients), five patients were independent walkers, one used a stroller as she did not crawl or walk yet at age 22 months, two used a manual wheelchair for outdoor mobility, and three used a motorized chair for outdoor mobility. Of the five children who ambulated without a mobility aid, four wore knee ankle foot orthoses and two walked with a crouched gait. Scoliosis was present in all but one patient who was only nine months of age at the last assessment. Four patients had malignant curves that had developed either during infancy or possibly prenatally (prenatal scoliosis). The level of the major curve was at the Scoliosis was present in all but one patient who was only nine months of age at the last assessment. Four patients had malignant curves that had developed either during infancy or possibly prenatally (prenatal scoliosis). The level of the major curve was at the thoracolumbar spine in two of these patients, the other two patients had left low thoracic curves. These curves all included the pelvis, causing moderate to severe pelvic obliquity(18–50◦ ). One patient had a mild but sweeping kyphosis including the pelvis, two had apical mid thoracic hyperkyphosis, whereas the last had thoracic hypokyphosis but with a thoracolumbar kyphosis. All four patients with malignant curves between 70◦ and 120◦ underwent non-fusion spinal instrumentation at 2, 4, 7 and 10 years of age in which expandable implants were placed to help control the curve during growth. Spine X-rays show a malignant curve in Figure 4. *Genes* **2021**, *12*, 1220 7 of 14 a thoracolumbar kyphosis. All four patients with malignant curves between 70° and 120° underwent non-fusion spinal instrumentation at 2, 4, 7 and 10 years of age in which expandable implants were placed to help control the curve during growth. Spine X-rays show a malignant curve in Figure 4.

thoracolumbar spine in two of these patients, the other two patients had left low thoracic

Characteristic phenotypic features of Escobar were observed in all 12 patients, including webbing over the neck, axillary, elbow, knee and/or fingers (Table 1). Downslanting palpebral fissures were observed in 10 patients (Table 2). Three patients had cleft palate and another five had high palate. Four patients had both posterior rotated ears and low-set ears, whereas another three patients either had posterior rotated ears or low-set ears. Figures 2 and 3 showcase pterygia and characteristic features in MPS among two

**Figure 4.** Patient 1 has compound heterozygous pathogenic *MYH3* variants. Curve was first noted at 13 months of age. (**A**,**B**) Posterior-anterior and lateral spine radiographs at 34 months of age showing a 68° curve with mild pelvic obliquity, and mild thoracolumbar kyphosis. (**C**) At 5 years old, the curve is 94°, with 33° pelvic obliquity and (**D**) more pronounced low thoracic kyphosis extending into the lumbar spine. (**E**) Latest follow-up at 7 years old, following halo gravity traction and a non-fusion spinal instrumentation at age 5. **Figure 4.** Patient 1 has compound heterozygous pathogenic *MYH3* variants. Curve was first noted at 13 months of age. (**A**,**B**) Posterior-anterior and lateral spine radiographs at 34 months of age showing a 68◦ curve with mild pelvic obliquity, and mild thoracolumbar kyphosis. (**C**) At 5 years old, the curve is 94◦ , with 33◦ pelvic obliquity and (**D**) more pronounced low thoracic kyphosis extending into the lumbar spine. (**E**) Latest follow-up at 7 years old, following halo gravity traction and a non-fusion spinal instrumentation at age 5.

> Severe curves were noted in three patients, with curves progressing more slowly than the malignant curves, so that surgery could be delayed. Two patients had radiographic findings of curves at 3 and 11 months of age, whereas the third had periodic radiographs which did not demonstrate a curve until 6 years of age. All three patients had a right thoracic curve pattern, with two having thoracic hypokyphosis, which was confluent with the lumbar lordosis, the third had a thoracolumbar kyphosis. One patient was undergoing brace treatment, another had a non-fusion spinal instrumentation procedure at 11 years Severe curves were noted in three patients, with curves progressing more slowly than the malignant curves, so that surgery could be delayed. Two patients had radiographic findings of curves at 3 and 11 months of age, whereas the third had periodic radiographs which did not demonstrate a curve until 6 years of age. All three patients had a right thoracic curve pattern, with two having thoracic hypokyphosis, which was confluent with the lumbar lordosis, the third had a thoracolumbar kyphosis. One patient was undergoing

> **Figure 5.** Patient 3 has compound heterozygous *CHRNG* variants and severe curve behavior. Curve first noted at 6 years of age. (**A**) Standing posterior-anterior spine films at 6 years old demonstrate a 19° right-sided thoracic curve. (**B**,**C**) A 71° curve with a lordosis spanning the thoracic and lumbar

> Moderate curves were found in two patients, both detected during the first year of life. Both were left sided curves, one a thoracolumbar curve with hyperkyphosis and hypolordosis, the other a lumbar curve with thoracic hypokyphosis. One patient underwent serial casting at 2 years of age, while the curve of the other did not progress past 25° until 12 years of age, and therefore was not treated. Two patients, 5 and 8 years of age at last

of age, and the last had a formal spinal fusion at 12 years of age (Figure 5).

spine. (**D**) Spinal fusion at 13 years of age.

and a non-fusion spinal instrumentation at age 5.

*Genes* **2021**, *12*, 1220 7 of 14

**Figure 4.** Patient 1 has compound heterozygous pathogenic *MYH3* variants. Curve was first noted at 13 months of age. (**A**,**B**) Posterior-anterior and lateral spine radiographs at 34 months of age showing a 68° curve with mild pelvic obliquity, and mild thoracolumbar kyphosis. (**C**) At 5 years old, the curve is 94°, with 33° pelvic obliquity and (**D**) more pronounced low thoracic kyphosis extending into the lumbar spine. (**E**) Latest follow-up at 7 years old, following halo gravity traction

show a malignant curve in Figure 4.

a thoracolumbar kyphosis. All four patients with malignant curves between 70° and 120° underwent non-fusion spinal instrumentation at 2, 4, 7 and 10 years of age in which expandable implants were placed to help control the curve during growth. Spine X-rays

brace treatment, another had a non-fusion spinal instrumentation procedure at 11 years of age, and the last had a formal spinal fusion at 12 years of age (Figure 5). the lumbar lordosis, the third had a thoracolumbar kyphosis. One patient was undergoing brace treatment, another had a non-fusion spinal instrumentation procedure at 11 years of age, and the last had a formal spinal fusion at 12 years of age (Figure 5).

Severe curves were noted in three patients, with curves progressing more slowly than the malignant curves, so that surgery could be delayed. Two patients had radiographic findings of curves at 3 and 11 months of age, whereas the third had periodic radiographs which did not demonstrate a curve until 6 years of age. All three patients had a right thoracic curve pattern, with two having thoracic hypokyphosis, which was confluent with

**Figure 5.** Patient 3 has compound heterozygous *CHRNG* variants and severe curve behavior. Curve first noted at 6 years of age. (**A**) Standing posterior-anterior spine films at 6 years old demonstrate a 19° right-sided thoracic curve. (**B**,**C**) A 71° curve with a lordosis spanning the thoracic and lumbar spine. (**D**) Spinal fusion at 13 years of age. **Figure 5.** Patient 3 has compound heterozygous *CHRNG* variants and severe curve behavior. Curve first noted at 6 years of age. (**A**) Standing posterior-anterior spine films at 6 years old demonstrate a 19◦ right-sided thoracic curve. (**B**,**C**) A 71◦ curve with a lordosis spanning the thoracic and lumbar spine. (**D**) Spinal fusion at 13 years of age.

Moderate curves were found in two patients, both detected during the first year of life. Both were left sided curves, one a thoracolumbar curve with hyperkyphosis and hypolordosis, the other a lumbar curve with thoracic hypokyphosis. One patient underwent serial casting at 2 years of age, while the curve of the other did not progress past 25° until 12 years of age, and therefore was not treated. Two patients, 5 and 8 years of age at last Moderate curves were found in two patients, both detected during the first year of life. Both were left sided curves, one a thoracolumbar curve with hyperkyphosis and hypolordosis, the other a lumbar curve with thoracic hypokyphosis. One patient underwent serial casting at 2 years of age, while the curve of the other did not progress past 25◦ until 12 years of age, and therefore was not treated. Two patients, 5 and 8 years of age at last follow up, had mild curves. They both had a forward lean on the standing sagittal radiographs with flexed hips and mild lumbar hyperlordosis. Figure 6 shows spine X-rays of a child with a moderate curve.

Thus, a total of five patients had undergone a non-fusion spinal instrumentation procedure. Interestingly, post-operative follow-up demonstrated that each of these patients had spontaneous fusions of vertebral levels, also called autofusions, in at least one uninstrumented vertebral interspace, depicted in Figure 7.

### *3.2. Molecular Analysis*

Whole genome or whole exome sequencing was performed in 8 of the 12 patients and their parents, when available. Parents were clinically unaffected, except in family F6, as described earlier. Pathogenic recessive variants in *CHRNG* were found in five patients, all of whom had compound heterozygous variants (Table 4). Among the five different *CHRNG* variants found in these individuals, four led to premature termination codons and one represented an in-frame duplication (p.Trp98\_Leu100dup) that had been described before [11]. According to family history, one other patient (P4) had compound heterozygous *CHRNG* mutations, but no detailed information was available and no DNA sample could be obtained.

Pathogenic variants in *MYH3* were found in two patients. One patient had a dominant de novo missense variant (p.Leu1204Pro), which affects the tail region of *MYH3*. This variant is not present in gnomAD and is predicted to be pathogenic by nine different prediction algorithms. Two other missense variants leading to substitutions of amino acids in the *MYH3* tail domain by proline residues have been described in the literature [12]. The other patient with pathogenic *MYH3* variants was compound heterozygous for a known recurrent splice variant (c.-9+1G > A) [12] and a novel missense variant (p.Ala183Pro). This missense is not present in gnomAD and affects the head region of MYH3, a domain where pathogenic missense variants are frequently observed [12].

A number of novel variants in genes reported to be associated with AMC were identified in patient P7, whose mother and maternal aunt have features of Escobar and scoliosis. Further validation of these variants is required given the novelty, and thus were not reported in this paper. follow up, had mild curves. They both had a forward lean on the standing sagittal radiographs with flexed hips and mild lumbar hyperlordosis. Figure 6 shows spine X-rays of a child with a moderate curve. child with a moderate curve. **C D**

follow up, had mild curves. They both had a forward lean on the standing sagittal radiographs with flexed hips and mild lumbar hyperlordosis. Figure 6 shows spine X-rays of a

*Genes* **2021**, *12*, 1220 8 of 14

**Figure 6.** Patient 10 has compound heterozygous *CHRNG* variants, and a moderate scoliosis, first identified during infancy. (**A**) Posterior-anterior supine film at 10 years old, demonstrating 15° right thoracic and 18° left lumbar curves, and lateral view (**B**) Showing mild thoracic hypokyphosis and hip flexion contractures driving lumbar hyperlordosis. At this size, the curves would be considered mild. (**C**,**D**) Posterior-anterior and lateral sitting spine films at 12 years of age, showing that the curve had mildly progressed to 28°, reaching criteria for a moderate curve. The sagittal profile continues to show mild thoracic hypokyphosis. **Figure 6.** Patient 10 has compound heterozygous *CHRNG* variants, and a moderate scoliosis, first identified during infancy. (**A**) Posterior-anterior supine film at 10 years old, demonstrating 15◦ right thoracic and 18◦ left lumbar curves, and lateral view (**B**) Showing mild thoracic hypokyphosis and hip flexion contractures driving lumbar hyperlordosis. At this size, the curves would be considered mild. (**C**,**D**) Posterior-anterior and lateral sitting spine films at 12 years of age, showing that the curve had mildly progressed to 28◦ , reaching criteria for a moderate curve. The sagittal profile continues to show mild thoracic hypokyphosis. curve had mildly progressed to 28°, reaching criteria for a moderate curve. The sagittal profile continues to show mild thoracic hypokyphosis. Thus, a total of five patients had undergone a non-fusion spinal instrumentation procedure. Interestingly, post-operative follow-up demonstrated that each of these patients had spontaneous fusions of vertebral levels, also called autofusions, in at least one uninstrumented vertebral interspace, depicted in Figure 7.

**Figure 7.** Patient 5 with compound heterozygous *CHRNG* variants, and severe scoliosis, first identified at 3 months of age. Spine bracing started at 24 months old. (**A**) Posterior-anterior and lateral standing films at 7 years old demonstrate a 53° right low thoracic curve, with (**B**) thoracic and lumbar lordosis on the lateral film, and flexed hips indicating hip flexion contractures. (**C**) Sitting spine films at 11 years old with curve at 66°, before non-fusion spinal instrumentation. (**D**) Autofusion along the length of the instrumented spine is seen 3 years after rod replacement at 17 years of age. **Figure 7.** Patient 5 with compound heterozygous *CHRNG* variants, and severe scoliosis, first identified at 3 months of age. Spine bracing started at 24 months old. (**A**) Posterior-anterior and lateral standing films at 7 years old demonstrate a 53◦ right low thoracic curve, with (**B**) thoracic and lumbar lordosis on the lateral film, and flexed hips indicating hip flexion contractures. (**C**) Sitting spine films at 11 years old with curve at 66◦ , before non-fusion spinal instrumentation. (**D**) Autofusion along the length of the instrumented spine is seen 3 years after rod replacement at 17 years of age.

**Figure 7.** Patient 5 with compound heterozygous *CHRNG* variants, and severe scoliosis, first identified at 3 months of age. Spine bracing started at 24 months old. (**A**) Posterior-anterior and lateral standing films at 7 years old demonstrate a 53° right low thoracic curve, with (**B**) thoracic and lumbar lordosis on the lateral film, and flexed hips indicating hip flexion contractures. (**C**) Sitting spine

Whole genome or whole exome sequencing was performed in 8 of the 12 patients and their parents, when available. Parents were clinically unaffected, except in family F6, as described earlier. Pathogenic recessive variants in *CHRNG* were found in five patients, all

along the length of the instrumented spine is seen 3 years after rod replacement at 17 years of age.

their parents, when available. Parents were clinically unaffected, except in family F6, as described earlier. Pathogenic recessive variants in *CHRNG* were found in five patients, all of whom had compound heterozygous variants (Table 4). Among the five different

Whole genome or whole exome sequencing was performed in 8 of the 12 patients and

*3.2. Molecular Analysis* 

*3.2. Molecular Analysis* 


**Table 4.** Genotype information.

WES: whole exome sequencing; WGS: whole genome sequencing. \* According to family history, detailed information not available.

### **4. Discussion**

Here we describe 12 individuals with variable pterygia, mild to severe flexion contractures of several joints and spine anomalies. In six of the patients, the disorder was caused by biallelic *CHRNG* mutations, one patient had an apparently dominant de novo mutation in *MYH3* and one patient was compound heterozygous for *MYH3* variants. Scoliosis was highly prevalent in our patient cohort and extremely severe in some. Intriguingly, all patients undergoing non-fusion spinal instrumentation subsequently developed fusion in at least one vertebral segment that had not been directly touched by the surgical intervention.

Biallelic loss of function variants in *CHRNG* are a well-established cause of Escobar syndrome [3,4]. Similar to previous studies, we observed that the recurrent frameshift c.459dup mutations in several unrelated patients [9]. In addition, we found that one of our patients had compound heterozygous *MYH3* variants, including the previously characterized splice variant c.-9+1G > A in the 50 untranslated region of the gene [12,13]. Interestingly the MYH3 c.-9+1G > A variant was initially described in individuals with spondylocarpotarsal synostosis syndrome, but the reported phenotype also included webbing, contractures and scoliosis [12] and thus had considerable overlap with the clinical characteristics of our patients. Moreover, *MYH3* mutations are associated with distal arthrogryposis type 8. Based on recommendations by Biesecker et al. [14] it might be useful to use the broad term *MYH3*-related disorder to encompass these different conditions.

Multiple pterygia are seen in many types of lethal forms of AMC [15–18], but only three genes have been associated with non-lethal MPS or Escobar syndrome, namely *CHRNG*, *MYH3*, and *TPM2*. One explanation might be the exclusive or predominant expression of these genes during development and their lack or diminished expression postnatally, whereas other genes seem to be pre- and postnatally expressed. Indeed, the *CHRNG* encoded gamma-subunit of AchR, as a developmental subunit, stops being expressed at the end of the second and start of the third trimester [19], a timepoint from which the postnatal/adult CHRNE-encoded epsilon subunit is incorporated into the AchR. That most certainly also explains why individuals with *CHRNG*-related MPS do not present with

clinical or electromyographic myasthenic symptoms after birth. In the same line, the *MYH3* encoded embryonic myosin heavy chain, despite its role on fiber type, fiber number, and muscle fiber differentiation [20], is much less expressed in human postnatal muscle [21]. This holds also true for *TPM2* [22].

The development of multiple pterygia probably reflects the spectrum of the most severe consequence and earliest onset of fetal akinesia. In this regard, pathogenic variants in *CHRNG*, *MYH3* and *TPM2* described to date have been associated with loss of function mutations. In *CHRNG*, nonsense, splice site and missense mutations alike have been shown to abolish AchR expression at the sarcolemma [3,4], thus probably completely abolishing neuromuscular transmission during embryonic and fetal development. Of note, the AchR is still expressed even in case of loss of function mutations in other genes involved in the maintenance and function of the neuromuscular junction, such as *RAPSN* or *MUSK* [9,23,24], leading to a severe fetal akinesia deformation sequence phenotype, but without pterygia.

Scoliosis in children with Escobar syndrome is extremely prevalent, with most curves eventually progressing to treatment. Other authors have reported prevalence rates of scoliosis in patients with multiple pterygium syndrome between 32% and 93% [25–28]. Of the 12 patients covered in this report, only three did not yet require any scoliosis treatment, one of whom is only 9 months old, the other two are under 10 years. Four of the patients had a spine deformity that behaved beyond what is typically considered severe, therefore we named curves that were greater than 50◦ before 4 years of age "malignant". The patients with malignant curves also had more severe limb involvement, most commonly including severe hip and knee flexion contractures with pterygium. Most malignant curves were first detected in infancy, some even in the neonatal time period, suggesting that some were actually prenatal curves, curves that reflected the patients' unchanging intrauterine position due to severe fetal akinesia. Importantly, other patients also had curves detected during their first year of life, but did not develop malignant curves, varying between mild and severe. The malignant curves all had pelvic obliquity near or greater than 20◦ , an indication of the uncompensated nature of these curves. Two of the patients in the severe category initially had their curves detected before their first birthday, but the third patient (Patient 3) was routinely monitored with serial radiographs, and no curve was detected prior to 6 years of age. All three of patients with severe curves underwent bracing of their spines, which allowed further growth prior to needing surgical stabilization of their curves. One of the shortcomings of the classification system we used is that a patient may progress from one class to another as they grow and their curves progress, and that the classification was based on a very small group of subjects. It is likely that some of the mild and moderate curves will progress to severe prior to the patient reaching skeletal maturity. We felt, though, that it provided some structure with which to analyze and group the subjects.

We expect that larger studies will provide insights, allowing for improvements of the scoliosis classification system used in this manuscript. A molecular diagnosis was confirmed in seven of our patients, two of which were found to have a *MYH3* gene mutation; the remaining five had a *CHRNG* mutation. The two patients with *MYH3* mutations both had malignant curves. The two other patients with malignant curves both had *CHRNG* gene mutations, of which one, Patient 2, was the older sibling of Patient 3, with the exact same mutation, but Patient 3 had a severe scoliosis. We have two other sibling pairs that we could not include in this study, since they did not have complete data available. One brother–sister pair both had malignant curves that started in infancy, whereas the brother of the other brother–sister pair had a malignant curve that started in infancy yet his sister's curve did not become severe (crossing 50◦ ) until after 10 years of age. Curve patterns varied modestly in our cohort. Two of the malignant curves were thoracolumbar extending to the pelvis, one left-sided the other right, whereas the other two were low thoracic apex curves, also extending nearly to the pelvis, both left-sided. Three of the spines had thoracic hyperkyphosis, but the fourth had a lordotic thoracic spine with a mild thoracolumbar

kyphosis. The severe curves were right sided mid- or lower thoracic-apex curves, with relatively lordotic thoracic spines, two patients had a mild thoracolumbar kyphosis. The mild and moderate curves in general were a combination of right thoracic and/or left lumbar curves, and usually a lumbar hyperlordosis. This variability complements the findings of Margalit and colleagues [29], who found three of their nine patients had rightsided thoracolumbar curves, and the rest had left-sided thoracolumbar curves. They did not describe the patterns of sagittal appearance of their patients. Coalescence of vertebral levels the spine appears to be common in patients with Escobar syndrome, particularly in the severe and malignant curves. Margalit et al. [29] noted the same on pre-operative computer tomography (CT) of their patients.

In our cohort, it is clear that spontaneous vertebral fusions occur, particularly in the severe and malignant curves, best seen in the children undergoing non-fusion spinal instrumentations. In these spines, progressive intervertebral fusions are seen both anteriorly and posteriorly in the uninstrumented section of the spine, suggesting either that lack of intervertebral motion, or the distraction of the space, leads to the fusion. We did not identify vertebral abnormalities or lack of vertebral segmentation in the films of our patients under 2 years of age, although we did not have CT scans and details could be difficult to visualize on the films. Therefore, we were unable to resolve if any of the intervertebral fusions were congenital failure of segmentation, but we suspect that most, if not all, were due to postnatal spontaneous fusions. Clearly, patients with Escobar syndrome need to be carefully monitored for the development of scoliosis, and aggressively treated to postpone the need for surgical intervention. Joo et al. [28] noted a tethered cord or a syrinx in 4 of their 16 patients. Although only one of our patients needed detethering of their spinal cord (Patient 5), treating physicians need to be vigilant for such possibilities. Both bracing and serial spine casting appear to be beneficial to some extent in controlling the curve and allowing further growth for at least a few years. Patients with spine-induced pelvic obliquity, particularly those apparent in infancy, likely have malignant curves. This seems to be particularly true for patients with *MYH3* mutations underlying their Escobar syndrome. Parents need to be informed about the challenging nature of the curve, and that non-fusion spinal instrumentation will be needed to try to modulate the curve during growth. Patients undergoing non-fusion spinal instrumentation are likely to experience spontaneous fusions of their spine, which may limit the amount of expansion possible during the child's growing years. Spine balance must be a priority at the initial implantation of the expandable device, as a formal fusion may not be necessary due to the spontaneous fusions, so long as spine balance is satisfactory. Conversely, if the spine is not well balanced, a formal fusion after a non-fusion spinal instrumentation will be very challenging due to the fusions.

### **5. Conclusions**

In conclusion, we found a unique spine phenotype in these patients with MPS caused by *CHRNG* and *MYH3* mutations. More detailed characterization using 3D imaging may help further refine this spine phenotype in patients with MPS.

**Author Contributions:** Conceptualization, N.D.-O., K.D., K.S., P.F.G. and H.v.B.; methodology, N.D.-O., K.D., F.R., A.G.G., K.S., P.F.G. and H.v.B.; data acquisition, N.D.-O., K.D., F.R., G.B., T.N.B., A.G.G., R.H., X.L., K.S., P.F.G. and H.v.B.; analysis, N.D.-O., K.D., F.R., G.B., T.N.B., A.G.G., R.H., X.L., K.S., P.F.G. and H.v.B.; resources, N.D.-O., K.D., F.R., G.B., T.N.B., A.G.G., R.H., X.L., K.S., P.F.G. and H.v.B.; writing—original draft preparation, N.D.-O., K.D., T.N.B., A.G.G., X.L., K.S., P.F.G. and H.v.B.; writing—review and editing, N.D.-O., K.D., F.R., G.B., T.N.B., A.G.G., R.H., X.L., K.S., P.F.G. and H.v.B.; funding acquisition, N.D.-O., K.D., F.R., R.H., P.F.G. and H.v.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Shriners Hospitals for Children for the pediatric arthrogryposis multiplex congenita registry at the Montreal and Philadelphia hospitals (grant 79150) to N.D.-O., R.H., F.R., G.B., P.F.G. and H.v.B., and from the Direction pour la Recherche et l'Innovation (DRCI) of the Grenoble Alpes University Hospital Center (Pediatric and Adult Registry for Arthrogryposis

Multiplex Congenita—project PARART) to K.D. Funding was obtained from the Research reported in this publication was also supported in part by the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under Award Number R03HD099516 to P.F.G. N.D.-O. holds a clinical research scholar award from the Fonds de la Recherche en Santé du Québec. We gratefully acknowledge the support of the Malika Ray, Asok K. Ray, M.D., FRCS/(Edin) Initiative for Child Health Research.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of McGill University's Faculty of Medicine, Canada (protocol A08-M30-19B on 4 October 2019), WCG Institutional Review Board, USA (protocol CAN1903 on 30 November 2019) and by the Institutional Review Board of Grenoble Alpes University Hospital Center, France (protocol 2205066v0 on 16 December 2019).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this paper.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding authors. The data are not publicly available due to the ethical and privacy nature of the data.

**Acknowledgments:** The authors are grateful to Vasiliki Betty Darsaklis from the Shriners Hospital for Children in Montreal, Canada for her assistance with data extraction and the Molecular Biology Facility of the Grenoble University Hospital, France for technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

### **References**


### **Shengru Wang 1,2,†, Xiran Chai 1,3,†, Zihui Yan 1,2,3,4,†, Sen Zhao 1,2,3,4, Yang Yang 1,2, Xiaoxin Li 2,4,5 , Yuchen Niu 2,4,5, Guanfeng Lin 1,2, Zhe Su 1,2, Zhihong Wu 1,2,4,5, Terry Jianguo Zhang 1,2,4,\* ,‡ and Nan Wu 1,2,4**


**Abstract:** *FGFR1* encodes a transmembrane cytokine receptor, which is involved in the early development of the human embryo and plays an important role in gastrulation, organ specification and patterning of various tissues. Pathogenic *FGFR1* variants have been associated with Kallmann syndrome and hypogonadotropic hypogonadism. In our congenital scoliosis (CS) patient series of 424 sporadic CS patients under the framework of the Deciphering disorders Involving Scoliosis and COmorbidities (DISCO) study, we identified four unrelated patients harboring *FGFR1* variants, including one frameshift and three missense variants. These variants were predicted to be deleterious by in silico prediction and conservation analysis. Signaling activities and expression levels of the mutated protein were evaluated in vitro and compared to that of the wild type (WT) *FGFR1*. As a result, the overall protein expressions of c.2334dupC, c.2339T>C and c.1261A>G were reduced to 43.9%, 63.4% and 77.4%, respectively. By the reporter gene assay, we observed significantly reduced activity for c.2334dupC, c.2339T>C and c.1261A>G, indicating the diminished FGFR1 signaling pathway. In conclusion, *FGFR1* variants identified in our patients led to only mild disruption to protein function, caused milder skeletal and cardiac phenotypes than those reported previously.

**Keywords:** *FGFR1* (Fibroblast growth factor receptor 1); genetic variations; congenital scoliosis

### **1. Introduction**

The Fibroblast growth factor receptor 1 (*FGFR1*) gene encodes a transmembrane cytokine receptor, which comprises an extracellular region of three immunoglobulin-like domains (D1, D2 and D3), a transmembrane helix and a cytoplasmic tyrosine kinase domain [1]. Although different isoforms have different tissue expression and varied affinity to FGFs, *FGFR1-IIIc*, spliced through the use of exon 8B, is the predominant isoform that carries out most of the functions of the *FGFR1* gene [2].

The downstream signaling of *FGFR1* is activated by the dimerization and activation of the receptor and autophosphorylation of the tyrosine kinase domains. These downstream signaling pathways include the mitogen activated protein kinases (MAPK), the

**Citation:** Wang, S.; Chai, X.; Yan, Z.; Zhao, S.; Yang, Y.; Li, X.; Niu, Y.; Lin, G.; Su, Z.; Wu, Z.; et al. Novel *FGFR1* Variants Are Associated with Congenital Scoliosis. *Genes* **2021**, *12*, 1126. https://doi.org/ 10.3390/genes12081126

Academic Editors: Roel Ophoff, Nancy Hadley-Miller, Philip F. Giampietro and Cathy L. Raggio

Received: 1 April 2021 Accepted: 14 July 2021 Published: 24 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

phosphatidylinositide 3 kinase/AKT (PI3K/AKT) and the phospholipase C γ (PLC) [1,3]. *FGFR1*-related signaling pathways are involved in the early development of the human embryo, and thus play an important role in gastrulation, organ specification and patterning of many tissues [4].

Many *FGFR1* mutations have been identified in both Kallmann syndrome and isolated hypogonadotropic hypogonadism (IHH) [5–9]. *FGFR1* loss-of-function mutations were also reported to be found in Kallmann syndrome patients with skeletal phenotypes, including oligodactyly, hemivertebrae and butterfly vertebrae [10] and *FGFR1* signaling was reported to be important for different stages of osteoblast maturation [11]. Mice models with *FGFR1* variants presented various skeletal phenotypes, especially vertebral malformation from cervical vertebrae to lumbar vertebrae, making *FGFR1* a candidate gene for congenital scoliosis [12]. However, whether *FGFR1* is associated with vertebral malformations in human remains unknown.

In this study, we analyzed variants of *FGFR1* identified in a cohort of congenital scoliosis (CS) and performed in vitro experiments to determine the effects of these variants on the protein function.

### **2. Materials and Methods**

### *2.1. Human Subjects*

A total of 424 sporadic Han Chinese probands who received a diagnosis of congenital scoliosis (CS) were consecutively collected into the cohort between 2009 and 2018 at Peking Union Medical College Hospital (PUMCH) under the the framework of the Deciphering disorders Involving Scoliosis and COmorbidities (DISCO, http://discostudy.org/, accessed on 15 March 2021) study. Demographic information, physical examination results, clinical symptoms on presentation, and a detailed medical history were obtained from each proband. Clinical diagnoses were confirmed by radiology imaging. Blood was obtained from all the probands and whole exome sequencing (WES) was performed.

A total of 942 Han Chinese individuals without evidence of congenital scoliosis or other congenital malformations from the DISCO project served as in-house controls. All inhouse controls provided their blood for DNA analysis and signed written informed consent.

### *2.2. Bioinformatic Analysis and Mutation Interpretation*

WES data processing was performed using the Peking Union Medical college hospital Pipeline (PUMP) [13,14] developed in-house. Computational prediction tools (Genomic Evolutionary Rate Profiling [GERP] [15], Combined Annotation Dependent Depletion [CADD PHRED-score, GRCh37-v1.6] [16], Sorting Intolerant Form Tolerant [SIFT] [17], Polyphen-2 [18], and MutationTaster [19]) were used to predict the conservation and pathogenicity of candidate variants. All variants were compared against population genomic databases such as the 1000 Genomes Project (http://www.internationalgenome.org/, accessed on 15 March 2021), the NHLBI GO Exome Sequencing Project (ESP) Exome Variant Server (http://evs.gs.washington.edu/EVS/, accessed on 15 March 2021) and the genome Aggregation Database (gnomAD, http://gnomad.broadinstitute.org/, accessed on 15 March 2021).

Candidate variants in *FGFR1* were extracted and filtered using the following criteria:


### *2.3. Site-Directed Mutagenesis*

Plasmids of pcDNA3.1+ with N-terminal myc-tagged WT and mutant *FGFR1c* cDNA (NM\_023110.2) were acquired from Beijing Hitrobio Biotechnology. The mutant constructs were sequenced on both strands to verify nucleotide changes.

### *2.4. Receptor Expression and Maturation Studies*

### 2.4.1. Endoglycosidase Digestion

Endoglycosidase assays were performed as previously published [8]. In brief, COS-7 cells (Cell Resource Center, Peking Union Medical College, Beijing, China) with 60–70% confluence were transiently transfected with 300 ng of plasmid containing myc-tagged WT or mutated *FGFR1* cDNA in 6-well plates using Lipofectamine 3000 reagent (Thermo Fisher Scientific, Waltham, MA, USA). Forty-eight hours post transfection, cells were washed with phosphate-buffered saline (PBS), and then, lysed with 100 µL of RIPA buffer (Thermo Fisher Scientific, Waltham, MA, USA) containing 1× protease inhibitor (Solarbio, Beijing, China). For deglycosylation analysis, all protein lysates were diluted to 10 µg/µL, and 9 µL of diluted lysate (90 µg of total protein) was subjected to PNGasef and EndoH digestion according to the manufacturer's recommendations (New England Biolab, Ipswich, MA, USA).

### 2.4.2. Western Analysis

Untreated or endoglycosidase-treated samples were resolved on gels under reducing conditions and then, subjected to Western analysis using an anti-myc primary antibody (clone 4A6, 1:1000, Upstate Biotechnology, Inc., Lake Placid, NY, USA) and a goat antimouse horseradish peroxidase-conjugated secondary antibody (1:5000, Bioss, Edinburgh, UK). Immunoreactivity was visualized using Western Lighting chemiluminescence reagent (Beyotime, Wuhan, China). To control for equal loading, blots were stripped using Restore Western Blot Stripping Buffer (Applygen Technologies, Beijing, China) and reprobed with horseradish peroxidase-conjugated anti-β-actin antibody (1:5000, Proteintech, Rosemont, IL, USA). FGFR1 and β-actin immunoreactivity were quantified by densitometry using an automatic chemiluminescence imaging system (Tanon, Shanghai, China). Overall expression levels of WT and mutant receptors were determined from the PNGase-treated samples and were normalized to their respective β-actin levels. The ratio between mutant and WT was reported. For receptor maturation studies, the upper (mature) and lower (immature) band densities were determined individually from the EndoH-treated samples, and the percent of mature fraction (maturation level) was calculated as overall protein divided by matured protein. The maturation levels of four variants were compared with the WT group, i.e., maturation ratio. Endoglycosidase and Western experiments were repeated three times.

### 2.4.3. FGF Reporter Gene Assay

The activation of downstream signaling pathways by wild type and mutated *FGFR1* constructs was interrogated using the luciferase-based reporter assay; the osteocalcin FGF response element (OCFRE) reporter reports the activity of the MAPK pathway downstream of FRS2α signaling [9]. In detail, L6 myoblasts (Cell Resource Center, Peking Union Medical College, Beijing, China), which are largely devoid of endogenous FGFRs and FGFs, were maintained in DMEM-H containing penicillin (100 U/L), streptomycin (100 ug/L), and 10% fetal calf serum. Transient transfections were performed at 60–70% cell confluency in 24-well plates with 300 ng of plasmid containing WT or mutant *FGFR1* cDNA, 400 ng of osteocalcin FGF response element-pGL3 plasmid, and 10 ng of pRL plasmid using Lipofectamine 3000 reagent (Thermo Fisher Scientific, Waltham, MA, USA). After 24 h of serum starvation, cells were treated for 16 h with FGF18 (10-8 M) in DMEM-H containing 0.1% BSA. The cells were lysed with passive lysis buffer (Promega, Madison, WI, USA), and assayed for luciferase activity using a Promega luciferase assay system. Experiments were performed in triplicate and repeated at least three times. Results of each experiment were normalized to the WT and the mean values of three experiments were calculated.

### 2.4.4. Statistical Analyses

The frequency of candidate variants of *FGFR1* was compared between the control group and the CS group using the Fisher Exact Test. Luciferase activities and overall

expression levels were normalized to WT (set as 100%) and mean values of mutant versus WT from all three experiments were compared using one-way ANOVA and Dunnett's multiple comparisons test. All charts were drawn and analyzed using GraphPad Prism 7 and *p* < 0.05 was considered significant for all analyses.

### **3. Results**

*3.1. Mutation and Phenotype Analyses*

In the 424 sporadic CS patients, 79 patients (18.6%) were found to have a molecular diagnosis by pathogenic genetic variants, as previously reported [13]. From the probands who remained undiagnosed, four likely deleterious heterozygous variants of *FGFR1*, including one frameshift variant and three rare missense variants (c.2334dupC; c.2339T>C; c.1107G>A; c.1261A>G), were identified (Table 1), presenting a significant mutational burden as compared with the in-house controls (one candidate variant in 942 control individuals, *p* = 0.035, Fisher Exact Test). The authenticity of all variants was validated by manual review of BAM files using the Integrative Genomics Viewer (http://igv.org, accessed on 15 March 2021).

**Table 1.** Demographic, phenotypic and variant information of four patients in our series. All variants' nomenclatures were based on the *FGFR1* transcript NM\_023110.2. All positions were aligned to GRCh37/hg19.


AF, allele frequency; pLI, probability of loss-of function intolerance; Ref, reference.

Patient #1 is a 13-year-old female with T6-10 segmentation defect (Figure 1a,b), fused left 9-10 ribs and mitral valve prolapse. She has a heterozygous duplication of nucleotide 2334 (c.2334dupC; p.Ser779GlnfsTer21). The variant was mapped in the intracellular region and post-translational phosphorylation site of the FGFR1 protein (Figure 2a) and was predicted by the NMD (nonsense-mediated decay) Prediction Tool (https://nmdpredictions. shinyapps.io/, accessed on 15 March 2021) to be located in the NMD-incompetent region (Figure 2b), suggesting that the variant is unlikely to cause nonsense-mediated decay. The variant was not found in population genomic databases, such as 1000G, ESP6500 and gnomAD. The CADD PHRED score of this variant is 32, indicating the deleteriousness of this variant.

**Figure 1.** Anteroposterior and lateral spinal X-ray of four patients: (**a**,**b**) Anteroposterior and lateral spinal X-ray of patient #1; (**c**,**d**) Anteroposterior and lateral spinal X-ray of patient #2; (**e**,**f**) Anteroposterior and lateral spinal X-ray of patient #3; (**g**,**h**) Anteroposterior and lateral spinal X-ray of patient #4.

Patient #2, a 4-year-old female, has T9 hemivertebrae, T8 butterfly vertebrae, and three ribs absent (Figure 1c,d). She has an *FGFR1* missense variant c.2339T>C (p.Phe780Ser). This variant was also mapped in the intracellular region and post-translational phosphorylation sites of the FGFR1 protein (Figure 2a). It was not found in most population databases, such as 1000 G, gnomAD and ESP6500. The variant was highly conservative across different vertebral species (Figure 2c). In silico prediction had contradictory results (tolerant or benign for SIFT, pathogenic for MutationTaster, Polyphen2, LRT and CADD PHRED-score).

Patient #3 is a male newborn affected with a T10 hemivertebrae (Figure 1e,f) with a missense variant c.1107G>A (p.Met369Ile). It is a novel mutation according to all population databases. In silico predictions were tolerant or benign for SIFT and Polyphen2, but deleterious for MutationTaster, LRT and CADD PHRED score.

**Figure 2.** Mapping and conservation analysis of four variants: (**a**) Mapping of four *FGFR1* variants revealed that c.2334dupC and c.2339T>C are located in the intracellular region and post-translational phosphorylation sites of the FGFR1 protein, whereas c.1261A>G is located in the transmembrane region and close to the post-translational phosphorylation site; (**b**) The result of NMD prediction of c.2334dupC showed that it is located in the NMD-incompetent region; (**c**) Mutation loci of the three missense variants (c.2339T>C, c.1261A>G and c.1107G>A) are highly conservative across different species.

Patient #4 is a male newborn who presents T10 hemivertebrae (Figure 1g,h). This patient has a novel missense variant (c.1261A>G; p.Ile421Val), which was mapped in the transmembrane region and close to the post-translational phosphorylation site of FGFR1 protein (Figure 2a). This variant is highly conservative among different vertebral species (Figure 2c). It was predicted to be deleterious by SIFT, MutationTaster, LRT and CADD PHRED score.

### *3.2. Functional Characterization of FGFR1 Variants*

### 3.2.1. Western Analysis

To identify the influences of these four variants on the function of the FGFR1 protein, we evaluated overall protein expression and maturation of the different *FGFR1* variants compared to WT. Endoglycosidase digestion and Western blotting analysis showed two immunoreactive-specific bands for WT FGFR1 at 140 and 120 kDa, corresponding to a differently N-glycosylated receptor. These two bands were reduced to a single lower molecular weight band following peptide N-glycosidase (PNGase) digestion to remove all types of N-linked carbohydrate chains. Treatment with endoglycosidase H (EndoH), which only removes high mannose N-linked sugars, merely affects the immature form (120 kDa), leaving the fully glycosylated mature form (140 kDa) intact. Thus, maturation rate can be calculated by dividing the band of 140 kDa from EndoH-treated samples into the band of 100 kDa. Overall expression level was quantified by measuring bands from PNGase-treated samples and normalized to the WT group (set as 100%). The overall expression of the frameshift variant was decreased to 43.9% compared with that of WT (*p* = 0.06), and those of three missense variants were reduced to 63.4% (*p* < 0.01), 82.8% (*p* = 0.887), and 77.4% (*p* = 0.743), respectively (Figure 3). As for maturation analysis, densitometric analysis revealed that 29.1% of the WT FGFR1 protein was expressed as a mature form (Figure 3). Consistent with our mapping analysis predicting that all four variants are not localized in the FGFR1 functional ectodomain, these mutant receptors showed no difference in the level of protein maturation, compared to WT (Figure 3).


**Figure 3.** Western blot analysis of COS-7 cells transiently transfected with WT or mutant *FGFR1* constructs reveal diminished protein expression levels of c.2334dupC, c.2339T>C, c.1107G>A and c.1261A>G. Overall expression was significantly decreased in all four variants, especially in c.2334dupC. No difference in protein maturation process was detected using a receptor deglycosylation. EV = empty vector, WT = wild type, UT = untreated, EH = EndoH-treated, PG = PNGase-f-treated.

### 3.2.2. FGF Reporter Gene Assay

To assess the influence of the four *FGFR1* variants on the receptor functionality, we first used the FGF-responsive reporter osteocalcin FGF response element-luciferase in L6 myoblasts, which acts downstream of the MAPK pathway (Figure 4). FGF18 is included in the FGF8 subfamily, which is expressed during somitogenesis and is essential for the morphogenesis of many tissues. In the *FGF18* knockout mice model, skeletal phenotypes have been detected [20], indicating an important role of FGF18 signaling in skeletal development. Previously, FGF18 was found to be expressed in and required for osteogenesis and chondrogenesis [21–25]. Compared to WT FGFR1, the receptor signaling capacity of the truncating variant (c.2334dupC) was reduced by 20.7% (*p* < 0.05, Figure 4). The responses of missense variants (c.2339T>C, c.1261A>G) were also significantly reduced by 26.6% and 28.8%, respectively (*p* < 0.01, Figure 4). These results indicated the diminished signaling pathway of FGFR1 activated by FGF18.

**Figure 4.** FGF reporter gene assay showing reduced signaling capacity of c.2334dupC, c.2339T>C and c.1261A>G. L6 myoblasts were transiently transfected with OCFRE-luciferase reporter together with wild type (WT) or mutant constructs and then treated with 10-8M FGF18. The average receptor signaling capacities of c.2334dupC, c.2339T>C and c.1261A>G were reduced by 20.7%, 26.6% and 28.8%, respectively. EV = empty vector. \* *p* value < 0.05, \*\* *p* value < 0.01.

### **4. Discussion**

In this study, we identified four pathogenic variants, namely one frameshift and three missense variations in patients with congenital scoliosis. The frameshift variant, c.2334dupC, found in a patient with vertebral segmentation defects and mitral valve prolapse, was the first *FGFR1* variant to be associated with spinal malformations and heart defects. Previous mouse models with *FGFR1* mutations were found to have malformations in both vertebrae and the heart [12], suggesting that *FGFR1* variants were associated with skeletal and cardiac abnormalities. Functional studies of this frameshift variant showed that this variant decreases overall protein expression compared with that of WT with a trend to significance but left protein maturation intact (Figure 3). The decreased overall protein expression of this variant might contribute to the diminished luciferase activity, suggesting a diminished signaling function induced by this variant (Figure 4). As the frameshift variant was located in the NMD-incompetent region, we proposed that this truncating variant did not lead to nonsense-mediated mRNA decay but only mildly affected the protein expression, and thus, merely resulted in mild skeletal and cardiac phenotypes.

As for the three missense variants, all of them were predicted to be deleterious by MutationTaster, LRT and CADD, but had different predictions by SIFT and Polyphen2. Two missense variants (c.2339T>C and c.1261A>G) were highly conservative across a wide range of vertebral species, suggesting them to be deleterious variants. Functional studies revealed that all missense variants had reduced overall protein expression, but only the decrease in c.2339T>C was statistically significant (Figure 3). Further luciferase assay indicated significantly reduced luciferase reporter activities (c.2339T>C and c.1261A>G), and thus, had diminished signaling functions (Figure 4). As the OCFRE reporter used in luciferase assays reports the activity of the MAPK pathway downstream of FRS2α signaling, we can conclude that c.2334dupC, c.2339T>C and c.1261A>G diminish the MAPK pathway.

Western blotting and maturation assay of the missense variant (c.1107G>A) showed a slightly decreased overall protein expression and normal maturation level. However, luciferase assay indicated that this variant has similar luciferase activity compared to WT, suggesting a normal effect on downstream signaling of this variant. As most proteins are redundant regarding their expression level, a minor decrease in expression level might not impact normal function. The missense variant c.1107G>A has an 82.8% expression level and a normal maturation ratio and thus, the matured protein of c.1107G>A is decreased to 84.6% compared to WT (82.8% times 102.17%), while matured protein of the other three variants is decreased to 48.8% for c.2334dupC, 59.6% for c.2339T>A and 69.7% for c.1261A>G compared to WT. Therefore, we proposed that matured protein with less than 70%~80% of WT could not be compensated by the redundant expression and might lead to the diminished signaling function indicated by the luciferase assay.

Furthermore, these three hypomorphic variants, including c.2334dupC, c.2339T>C and c.1261A>G, were mapped around post-translational modification sites and may affect protein phosphorylation, which plays an important role in normal protein function. Ying et al. [26] reported a patient with cryptorchidism, micropenis, strabismus, and hypopsia, who was diagnosed with nIHH. The patient had a de novo mutation in *FGFR1* (c.2008G>A), which induced a post-translational modification defect, including defective glycosylation and impaired trans-autophosphorylation. This study revealed the significance of posttranslational modification of FGFR1. Based on in silico analysis and functional study results, we believe these three hypomorphic variants (c.2334dupC, c.2339T>C and c.1261A>G) of *FGFR1* may be associated with spinal defects in our patients. As for patient #3 with c.1107G>A, we propose that his skeletal defects are caused by other unknown genetic or environmental factors.

Pathogenic loss-of-function variants of the *FGFR1* gene were reported to be involved in patients with Kallmann Syndrome, including hypogonadotropic hypogonadism and anosmia [5,10,27], and isolated HH [6–9]. Patients with *FGFR1* mutations also presented with skeletal phenotypes [7,9,10], including oligodactyly on both feet, fusion of metacarpal bones, hemivertebrae, butterfly vertebrae and split hand/foot malformation.

In our cohort, a broad range of skeletal phenotypes were observed, as one patient had failure of segmentation, one patient had mixed defects and two patients had failure of formation. This is consistent with previous studies of *FGFR1* pathogenic variants, in which patients with HH can present a varied spectrum of reproductive phenotypes and non-reproductive phenotypes [8,10]. Furthermore, different patients carrying identical *FGFR1* mutations were observed to exhibit largely variable expressivity of reproductive phenotypes [8]. *FGFR1* signaling is involved in the determination of mesodermal cell fates and regional patterning of the mesoderm during gastrulation [28], and thus, affects organ specification. For the *FGFR1* signaling pathway, different organ systems respond to ligand binding with discrepant patterns [29], and several distinct downstream pathways, such as Erk1/2, Frs2, Crk proteins and Plcγ, are involved [30]. Given the broad function of *FGFR1* in embryo development, wide crosslink with other signaling pathways, tissuespecific response patterns and different downstream pathways, it is reasonable that patients with *FGFR1* mutations can present distinct phenotypes affecting different organ systems. However, the detailed mechanisms through which *FGFR1* mutations lead to different diseases need to be further studied and clarified.

In previous studies, patients with different FGFR1 domains affected have been revealed to present different phenotype spectra. The variants found in these patients all impair the functional domain of FGFR1 protein, including exon 1U, which is located around multiple transcription factor-binding sites, the FRS2α-binding domain and the tyrosine kinase domain [7,9]. Compared to these studies, patients in our cohort only presented mild spine and heart defects. As none of our variants were located in the functional region of the FGFR1 protein or led to severe damage to protein structure, we hypothesized that mild variants in our patients can only result in mild phenotypes compared with those in previous studies.

### **5. Conclusions**

In conclusion, we found four *FGFR1* variants in our CS cohorts—one frameshift variant (c.2334dupC) and three missense variants (c.2339T>C; c.1107G>A; c.1261A>G). Functional studies revealed diminished signaling function and reduced protein expression in three of them (c.2334dupC; c.2339T>C; c.1261A>G). These variants in our patients only caused mild damage to the protein expression, and thus, resulted in mild skeletal and cardiac phenotypes, compared to those in previous studies.

**Author Contributions:** Conceptualization, S.W., X.C., Z.Y., N.W. and T.J.Z.; methodology, X.C., S.Z. and X.L.; software, S.Z. and X.L.; validation, S.W., S.Z., Y.N. and G.L.; formal analysis, X.C., Z.Y. and Z.S.; investigation, X.C.; resources, S.W. and Y.N.; data curation, X.C., Z.Y. and S.Z.; writing original draft preparation, X.C., S.W., Z.Y; writing—review and editing, Z.Y., S.Z., G.L., Y.Y. and N.W.; visualization, X.C.; supervision, Y.Y., N.W., Z.W. and T.J.Z.; project administration, N.W., Z.W. and T.J.Z.; funding acquisition, N.W., Z.W. and T.J.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded in part by the National Natural Science Foundation of China (81902178 to S.W., 81972037 to J.Z., 81822030 and 82072391 to N.W., 81930068 and 81772299 to Z.W.), Beijing Natural Science Foundation (No. L192015 to J.Z., JQ20032 to N.W., 7191007 to Z.W.), Capital's Funds for Health Improvement and Research (2020-4-40114 to N.W.), Tsinghua University-Peking Union Medical College Hospital Initiative Scientific Research Program, Non-profit Central Research Institute Fund of Chinese Academy of Medical Sciences (No. 2019PT320025), and the PUMC Youth Fund & the Fundamental Research Funds for the Central Universities (No. 3332019021 to Shengru W.).

**Institutional Review Board Statement:** Approval for the study was obtained from the ethics committee at Peking Union Medical College Hospital (JS-098).

**Informed Consent Statement:** Written informed consent was provided by each participant.

**Data Availability Statement:** Data are available upon reasonable request. The datasets analyzed during the current study are available from the corresponding author on reasonable request.

**Acknowledgments:** We thank the patients and their families who participated in this research. We also thank the Nanjing Geneseq Technology Inc. for technical help in sequencing, and Ekitech Technology Inc. for technical support in database and data management. We also thank the Deciphering disorders Involving Scoliosis and Comorbidities study.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

