**Severe Peripheral Joint Laxity is a Distinctive Clinical Feature of Spondylodysplastic-Ehlers-Danlos Syndrome (EDS)-***B4GALT7* **and Spondylodysplastic-EDS-***B3GALT6*

**Stefano Giuseppe Cara**ffi **1, Ilenia Maini 1,2, Ivan Ivanovski 1,3, Marzia Pollazzon 1, Sara Giangiobbe 1, Maurizia Valli 4, Antonio Rossi 4, Silvia Sassi 5, Silvia Faccioli 5, Maja Di Rocco 6, Cinzia Magnani 7, Belinda Campos-Xavier 8, Sheila Unger 8, Andrea Superti-Furga 8,\* and Livia Garavelli 1,\***


Received: 13 August 2019; Accepted: 10 October 2019; Published: 12 October 2019

**Abstract:** Variations in genes encoding for the enzymes responsible for synthesizing the linker region of proteoglycans may result in recessive conditions known as "linkeropathies". The two phenotypes related to mutations in genes *B4GALT7* and *B3GALT6* (encoding for galactosyltransferase I and II respectively) are similar, characterized by short stature, hypotonia, joint hypermobility, skeletal features and a suggestive face with prominent forehead, thin soft tissue and prominent eyes. The most outstanding feature of these disorders is the combination of severe connective tissue involvement, often manifesting in newborns and infants, and skeletal dysplasia that becomes apparent during childhood. Here, we intend to more accurately define some of the clinical features of *B4GALT7* and *B3GALT6*-related conditions and underline the extreme hypermobility of distal joints and the soft, doughy skin on the hands and feet as features that may be useful as the first clues for a correct diagnosis.

**Keywords:** beta-1,3-galactosyltransferase 6 (*B3GALT6)*; beta-1,4-galactosyltransferase 7 (*B4GALT7)*; spondyloepimetaphyseal dysplasia with joint laxity (SEMDJL1; SEMDJL-Beighton type); Ehlers–Danlos syndrome (EDS); spondylodysplastic Ehlers–Danlos syndrome (spEDS); spEDS-*B3GALT6*; spEDS-*B4GALT7*; extreme laxity of distal joints; soft; doughy skin on the hands and feet

#### **1. Introduction**

The extracellular matrix (ECM) of connective tissues like cartilage, bone, tendon, ligaments and skin is a complex interacting arrangement of proteins, glycoproteins and proteoglycans (PGs) that forms the scaffold of the human body and also constitutes the vital environment supporting cell growth and differentiation. PGs are macromolecules that are complex in their structures and represent an important component of both the ECM and of cell membranes. PGs have mechanical and osmotic functions but are strongly involved in signaling pathways controlling cell-to-cell interactions as well as cell growth and differentiation and tissue repair [1–3]. PGs are composed of a core protein and one or more glycosaminoglycan (GAG) chains that are long polysaccharides consisting of repeating disaccharide units. On the basis of the composition of these units, the PGs are subdivided into heparan sulfate (HS) PGs and chondroitin sulfate (CS)/dermatan sulfate (DS) PGs [4]. The biosynthesis of the GAG side chain begins with the creation of a common tetrasaccharide, a so called "linker", covalently linked to the hydroxyl group of serine residues on the core protein (O-glycosylation).

The linker region is synthesized through a coordinated mechanism involving specific glycosyltransferases. It starts with the transfer of a xylose residue by xylosyltransferase I or II (encoded by *XYLT1*, OMIM \*608124, and *XYLT2*, OMIM \*608125), followed by the addition of two galactose residues by galactosyltransferase I, encoded by *B4GALT7* (OMIM 604327) and galactosyltransferase II encoded by *B3GALT6* (OMIM \*615291) (Figure 1). Subsequently there will be a transfer of glucuronic acid by glucuronosyltransferase I encoded by *B3GAT3* (OMIM \*606374). The so-called "linkeropathies" are recessive conditions caused by variations in genes encoding the enzymes responsible for the synthesis of the linker region of PGs.

**Figure 1. A**. Galactosyltransferases I (B4GALT7) and II (B3GALT6) catalyze the addition of galactose units at specific positions of the tetrasaccharide region linking Serine residues on the protein backbone to glycosaminoglycan (GAG) chains. Xyl = Xylose, Gal = Galactose, GlcA = Glucuronic Acid. **B**. Functional defects of B4GALT6 or B4GALT7 inhibit addition of one or both Gal monosaccharides, respectively, and prevent GAG chain polymerization.

Variants of the *XYLT1* gene are responsible for skeletal dysplasias, including Baratela–Scott syndrome (OMIM 615777, formerly erroneously called Desbuquois dysplasia type 2) [5–10], variants of *XYLT2* are the cause of the spondylo-ocular syndrome (OMIM 605822) [11–13], while variants of *B3GAT3* cause a Larsen-like syndrome and a more severe skeletal dysplasia (OMIM 245600) [14–20].

In particular, variants in the *B4GALT7* and *B3GALT6* genes are the cause of a combined skeletal and connective tissue phenotype.

A *B4GALT7*-related condition was initially described as "progeroid type 1" Ehlers–Danlos syndrome (EDS), due to the phenotype characterized by premature aging and loose elastic skin in one reported patient [21–30]. Subsequently, Cartault et al. in 2015 described a unique homozygous *B4GALT7* mutation, p.Arg270Cys, in a large cohort of patients affected by the so-called Larsen of Reunion Island syndrome [31]. This contributed to define the currently known phenotype of *B4GALT7*-related EDS, with round flat face, proptosis, short stature, hypotonia, radioulnar synostosis, osteopenia, hyperextensible skin and joint hypermobility (OMIM #130070) [24,25,31–33].

*B3GALT6* variants cause a single disorder that was alternatively described as spondyloepimetaphyseal dysplasia (SEMD) with joint hypermobility (SEMDJL1 or SEMDJL Beighton type; OMIM #271640) in the "Nosology and classification of genetic skeletal disorders: 2015 revision" [34–43] or as severe EDS-like disorder (OMIM #615349) because of the striking joint laxity and muscular hypotonia in infancy [44,45]. Affected patients generally present clinical features of connective tissue weakness in infancy, and subsequently develop signs of skeletal dysplasia; sometimes these signs are already present at birth. Some patients develop life-threatening complications, such as aortic dilatation, aneurysms and cervical spine instability.

Due to the mixed phenotype, with signs of EDS and signs of skeletal dysplasia, the *B4GALT7* and *B3GALT6*-related syndromes have been called "spondylodysplastic EDS" (spEDS) in the recently revised EDS classification of 2017 [46–48].

Hereby we present the clinical features of 3 patients with spEDS. Two are novel patients: Pt. 1 with *B4GALT7*-related EDS, and Pt. 3 with a severe *B3GALT6*-related EDS. Pt. 2 was originally reported as *B3GALT6*-related SEMDJL by Nakajima et al. in 2013, but is included here because his clinical description was largely unpublished and his condition has undergone a significant evolution. Through the description of these patients and a review of previous literature reports, we contribute to a more accurate definition of the clinical features associated with *B4GALT7*- and *B3GALT6*-related syndromes. In particular, we point out the extreme distal joint hypermobility and skin hyperextensibility (more evident in hands and feet) these two conditions have in common, and the peculiar radiological signs of each syndrome. The combination of these features is the main indicator for clinicians to suspect these ultra-rare conditions.

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

#### *2.1. Metabolic Labeling of Fibroblast Cultures and PG Synthesis Analysis*

Skin fibroblasts from the patient and controls were cultured in minimum essential medium (MEM) with 10% fetal calf serum and antibiotics at 37 ◦C in a humidified atmosphere containing 5% CO2. Confluent cells in 10-cm2 petri dishes were preincubated for 4 h with or without 1 mM p-nitrophenyl β-D-xylopyranoside in MEM containing 250 μM cold Na2SO4 without FCS in 5% CO2 at 37◦C. Cells were then labeled with 150 μCi/ml Na2[ 35S]O4 in the same medium for 24 h, as described previously [49]. At the end of the labeling period, an equal volume of 100 mM sodium acetate buffer, pH 5.8, containing 8 M urea, 4% Triton X-100, 20 mM ethylenediaminetetraacetic acid, 20 mM N-ethylmaleimide, and 1mM phenylmethanesulfonyl fluoride was added to the medium. The cell layer was harvested in 50 mM sodium acetate buffer, pH 5.8, containing 2 M urea, 2% Triton X-100, and an aliquot was used for protein content determination with the bicinchoninic acid Protein Assay (Pierce), while the remainder was added to the medium. Samples were loaded on 1 ml DEAE SephacelTM columns; after column washing with 50 mM sodium acetate buffer, pH 6.0, 8 M urea, 0.15 M NaCl, 0.5% Triton X-100 and proteinase inhibitors, PGs were eluted with 1 M NaCl in the same buffer, recovered by precipitation with 9 volumes of ethanol. The pellet was washed with 70% ethanol and then solubilized in water; PGs were quantified by measuring the 35S-activity using a liquid scintillation counter and normalized to the protein content.

Labeled PGs synthesized by cells in absence of p-nitrophenyl β-D-xylopiranoside and purified as described above, were lyophilized, dissolved in 4 M guanidinium chloride, 50 mM sodium acetate buffer, pH 6.0, 0.5% Triton X-100, and analyzed by Size Exclusion Chromatography. Samples were then loaded on a Superose 6 10/300GL column (GE Healthcare) and eluted in the same buffer at 0.2 ml/min. Fractions of 0.4 ml were collected and 35S-activity was measured by scintillation counting [50].

#### *2.2. Molecular Genetic Testing*

Genomic DNA of each patient was isolated from peripheral blood mononuclear cells and analyzed using a capture-based IonAmpliSeq custom panel on an Ion Torrent series S5 instrument (Thermo Fisher Scientific). Results were filtered for the *B4GALT7* and *B3GALT6* genes. Identified variants were confirmed by PCR amplification and subsequent bidirectional Sanger sequencing, using primers encompassing the entire exon and intron–exon boundaries comprising the variants in the case of *B4GALT7*, or the entire coding sequence and surrounding untranslated regions in the case of *B3GALT6* single exon. In parallel, DNA samples from the respective parents were analyzed by Sanger sequencing in order to evaluate parental segregation. Primer sequences are available upon request.

#### **3. Patients and Results**

#### *3.1. First Patient*

#### 3.1.1. Clinical Report

The boy is the first child of healthy, non-consanguineous parents. The pregnancy was unremarkable and the mother was not exposed to any known teratogens during pregnancy. Fetal movements were described as normal. He was born by cesarean delivery at week 38 with birth weight of 2520 g (3rd–10th centile), length 43 cm (<3rd centile), and head circumference of 32 cm (3rd–10th centile). Apgar scores were 9/10. He presented with mesomelic shortening of upper limbs, ulnar deviation of fingers and left hip dysplasia. Cerebral and renal ultrasound were normal, while echocardiography showed atrial septal defect. The baby was discharged from hospital at the age of 6 days, with diagnosis of "arthrogryposis with ulnar deviation of fingers, left hip dysplasia, and congenital heart disease", and later he was fitted with a finger extension device and was given pelvic bimalleolar hip plaster for left hip dysplasia. He was seen at our Medical Genetics Unit at 3 months of age (Figure 2): length and weight, calculated by taking into account the hip plaster, were respectively 51 cm (3rd–10th centile) and 4900 g (3rd centile), and head circumference was 41 cm (25th centile). The physical examination showed: round flat face, mild proptosis, slightly blue sclerae (Figure 2A–D), joint hypermobility especially evident in the hands, mesomelic shortening of the arms, abnormal prono-supination of the left upper limbs, often flexed, short hands and feet with soft skin, bilateral ulnar deviation of fingers, 2nd finger clinodactyly, 2nd–5th finger camptodactyly on the left, and 3rd and 4th finger camptodactyly on the right (Figure 2F–G).

During the clinical follow-up, we noticed an improvement in the finger alterations, but a worsening of the cutaneous hyperextensibility and joint hypermobility. At the age of 3 years and 7 months, he presented with round face, blue sclerae, proptosis (Figure 2D), soft hyperextensible skin and extreme joint hypermobility (Beighton score: 5) in particular of hands and feet, with significant improvement in finger ulnar deviation, clinodactyly, camptodactyly and overlapping toes (Figure 2H–K). Height and weight were 87 cm (<3rd centile) and 9920 g (<3rd centile), head circumference was 49 cm (10th centile). Span was 82 cm and span/height ratio was 0.94. Early psychomotor development was delayed at least in part owing to his orthopedic conditions. He gained head control at 3 months, was able to sit at 3 years and walk aided at the age of 4 years. He uttered his first words at 2 years, but his language skills improved rapidly and significantly thereafter: at 5 years, psychological evaluation reported an IQ of 105.

The skeletal X-rays demonstrated wide anterior ribs, bilateral bowed ulna and radius with dislocation/subluxation and radioulnar synostosis, bilateral metaphyseal widening of the radius, bilateral short and dysmorphic 2nd finger middle phalanx, short first metacarpal (Figure 3A–F).

**Figure 2.** Patient 1 **A**) Age 3 months: ulnar deviation of fingers; **B**) Age 3 months: round flat face, mild proptosis, mesomelic shortening of upper limbs, folded skin on forearm; **C**) Age 3 months: flat profile; **D, F–G**) Age 3 months: ulnar deviation of fingers, 2nd finger clinodactyly, 2nd–5th finger camptodactyly on the left hand, 3rd and 4th finger camptodactyly on the right hand; **E**) Age 3 years 7 months: round face, blue sclerae, proptosis; **H**–**K**) Age 3 years 7 months: soft, hyperextensible skin and extreme joint hypermobility (Beighton score: 5) in particular of hands and feet; significant improvement in finger ulnar deviation, clinodactyly, camptodactyly of the second finger, and overlapping toes.

**Figure 3.** Patient 1 **A**–**B**, **D**–**E**) Bilateral bowing of ulna and radius with dislocation/subluxation and radioulnar synostosis, metaphyseal widening of the radius; **C**) bilateral short and dysmorphic 2nd finger middle phalanx, short first metacarpal; **F**) hip dysplasia.

Electrencephalography, electromyography and brain magnetic resonance imaging were all normal. At the age of 5 years, molecular analysis of *B4GALT7* was performed, showing the presence of two novel pathogenic mutations, indicative of spEDS.

#### 3.1.2. Molecular and Functional Analysis

Karyotype was normal, 46,XY. Our first diagnostic hypothesis was either myopathy or collagen VI pathology, but molecular analysis of the *COL6A1*, *COL6A2*, *COLA613*, *RAPSN* genes did not reveal any rare or pathogenic variant. Subsequently, by analyzing a panel of EDS-related genes, two variants were identified in compound heterozygosity in exons 2 and 3 of the *B4GALT7* gene: NM\_007255:c.[277dupC];[628C>T], NP\_009186:p.[(His93Profs\*73)];[(His210Tyr)] (Figure 4).

**Figure 4. A**) Pedigree of Family 1; **B**) Schematic representation of the B4GALT7 protein: TM = transmembrane domain, GalT = Galactosyl-transferase catalytic domain. Protein variants reported in the literature are annotated in the cartoon (see Table 1 for references). Top half: missense variants (black). Bottom half: truncating variants (red).

The single nucleotide insertion c.277dupC, inherited from the mother, creates a new reading frame, terminating with a premature stop codon 73 residues downstream, and its transcript is predicted to undergo nonsense-mediated decay (https://nmdprediction.shinyapps.io/nmdescpredictor/). This variant was also described in another patient by Salter et al. in 2016 [32]. The heterozygous C>T transition at nucleotide 628, inherited from the father, results in a non-conservative Histidine to Tyrosine substitution at codon 210 and occurs in a highly conserved residue. This variant has not been reported before, but it is considered deleterious by multiple prediction algorithms (MutationTaster, PolyPhen2, SIFT) and is absent from reference population databases (gnomAD, 1000 Genomes). Based on these criteria and on the clinical features of the patient, we assume that the combination of these variants in trans is causative.

The study of PG synthesis on patient's fibroblasts obtained by skin biopsy did not reveal any significant differences compared to normal controls, neither at basal conditions nor in the presence of beta-xyloside to enhance glycosaminoglycan synthesis. However, the level of PG synthesis in the patient's fibroblasts was at the lower normal limit. Analysis through gel filtration to assess the fraction of whole PGs with respect to glycosaminoglycans did not show any differences. Although these results were not significant *per se*, the slightly reduced PG levels combined with the genetic testing offered some indication that the variants identified in *B4GALT7* may in fact have a deleterious effect on protein function.

**1.** Patients with *B4GALT7* variants: Review of the literature.

**Table**

#### *Genes* **2019**, *10*, 799



 for *B4GALT7* variants [46]. The 5 patients from Hernandez et al. (column 1) had clinical data compatible with spEDS-*B4GALT7*, but were not considered in the clinical signs totalsbecause it was unclear whether the clinical diagnosis was confirmed by molecular testing or not [21–23].

#### *3.2. Second Patient*

#### 3.2.1. Clinical Report

This patient was partially described in Nakajima et al., 2013 as P9 [45]. He is the first child of healthy non-consanguineous parents, born at the 40th week of gestation by vaginal delivery. During pregnancy, no known exposures to potential teratogens were detected. The mother reported normal fetal movements. Prenatal ultrasound showed bilateral megaureter. At birth, weight was 2870 g (10th centile), length was 48 cm (3rd–10th centile) and head circumference measured 34 cm (25th centile). He presented with: joint hypermobility, soft skin, bilateral congenital talipes equinovarus and right hip dysplasia, which was treated with a Pavlik harness. Renal ultrasound showed a mild bilateral caliceal dilatation, while renal scintigraphy was normal. During the first year of life he was diagnosed with early-onset scoliosis, and he began to wear an orthopedic corset from the age of 4 years. Two years later. he was operated for scoliosis after correction with Halo Gravity Traction and he subsequently received three operations for the lengthening of the pins (until the age of 10 years). He suffered from frequent fractures: a fracture of the left femur after a low trauma at the age of 5 years and 6 months and to the right femur at the age of 8 years and 8 months and a micro-fracture of the right proximal fibula in an area of reduced bone density, compatible with an osseous cyst, at the age of 10 years. Furthermore, he had recurrent luxation of the toes.

Early psychomotor development was normal, but he later suffered from motor delay due to his skeletal and large joint anomalies. He gained head control at the age of 2 months and could sit up at the age of 8 months. He could walk unaided at the age of 2 years and uttered his first words at the age of 1 year. He currently attends the first year of middle school with a support teacher owing to a diminished ability to concentrate. He does not present with intellectual disability or other behavioral disturbances.

At the age of 12 years and 7 months (Figure 5), height was 109cm (<3rd centile <-6SD), weight 29 kg (<3rd centile), head circumference 54 cm (50th–75th centile), span 114 cm, span/height ratio 1.06. Pubertal stage was A0P1B1, with testicular volume 1–2 ml on the left side and 2–3ml on the right. He had sparse hair, high and prominent forehead, large ears, sparse eyebrows, large deeply-set eyes, blue-gray sclerae, hypoplastic columella and misalignment of teeth, which were small, with abnormal enamel and yellow-brown discoloration (Figure 5A–C). He had thin, pale, extremely soft skin, with a prominent venous patterning on the trunk and limbs, limited elbow extension, skin hyperextensibility and distal joint hypermobility especially of the hands (Beighton scale: 6), long and tapered fingers, normal nails. His feet had hypoplastic nails, short and overlapping toes, hallux valgus (Figure 5D–H).

The skeletal X-rays demonstrated severe kyphoscoliosis, osteopenia, thin metacarpals and phalanges (Figure 6A–H).

Audiometric and ophthalmological evaluations were normal. Abdominal ultrasound showed ptosic kidneys, bilateral pelvic ureteral dilatation and thickening of the bladder walls. Echocardiography demonstrated mild dilatation of the aortic bulb and mild mitral valve prolapse.

**Figure 5.** Patient 2 **A**–**C**) Sparse hair, high and prominent forehead, sparse eyebrows, deeply-set eyes, blue sclerae, hypoplastic columella, large ears; **D**–**H**) Thin, pale, extremely soft skin, with prominent veins on the trunk and limbs, limited elbow extension, skin hyperextensibility and distal joint hypermobility especially of the hands, long and tapered fingers. Feet: hypoplastic nails, short, overlapping toes, hallux valgus.

**Figure 6.** Patient 2 **A**–**H**) Severe, early onset kyphoscoliosis, osteopenia, thin metacarpals and phalanges.

#### 3.2.2. Molecular Analysis

Sanger sequencing confirmed the two previously reported variants in the single exon of the *B3GALT6* gene [45]: NM\_080605:c.[353delA];[925T>A], NP\_542172:p.[(Asp118Alafs\*160)];[(Ser309Thr)] (Figure 7A,B).

The frameshift variant was inherited from the father and the missense variant from the mother. Both variants can be considered pathogenic, according to the predicted effect on the protein, the consistency of the clinical phenotypes, and their rarity in reference population databases such as gnomAD (MAF=0.0026% and absent, respectively). Variant c.353delA has also been found in our third patient (reported below), while variant c.925T>A had already been reported in other patients in the literature [45,48]. Since this gene is encoded by a single exon, its frameshift variants, identified in various patients, are expected to escape nonsense-mediated mRNA decay. They probably exert a pathogenic effect by leading to an unstable transcript or an unstable/non-functional protein, as suggested by experimental evidence [48].

**Figure 7. A**) Pedigree of Families 2 and 3; **B**) Schematic representation of the B3GALT6 protein: TM = transmembrane domain, GalT = Galactosyl-transferase catalytic domain. Protein variants reported in the literature are annotated in the cartoon (see Table 2 for references). Top half: missense variants (black). Bottom half: frameshift variants (red), in-frame indels (green), other length-altering variants (blue); **C**) Protein projection of the two alleles from Pt.3; dashed boxes indicate an altered or extended reading frame.

#### *3.3. Third Patient*

#### 3.3.1. Clinical Report

He is the second child of non-consanguineous parents, born at 35th week of gestation by caesarean section, owing to the premature rupture of the amniotic sac. During pregnancy, no known exposures to potential teratogens were detected. Prenatal ultrasound indicated bilateral talipes equinovarus. The mother reported normal fetal movements. At birth, weight was 2500 g (50th centile), length was 47 cm (50th–75th centile) and head circumference measured 34 cm (90th centile). Apgar scores were 1':7 5':9. He presented with joint hypermobility, soft skin, adducted thumbs, right talipes equinovarus. In the second year of life he was diagnosed with early-onset scoliosis, and he began to wear an orthopedic corset.

At age 6 years, he was operated for bilateral cryptorchidism and at age 9 years, he underwent a gastrostomy to correct Barrett's oesophagus.

His motor development was delayed. He gained head control at the age of 4–5 months and could sit up at the age of 24 months. He could walk aided at the age of 5 years, he was able to walk alone at the age of 6 years and 6 months and uttered his first words at the age of 1 year. He currently attends the second year of middle school with a support teacher owing to a diminished ability to concentrate. He does not present with intellectual disability or other behavioral disturbances.

Examination at the age of 12 years and 11 months showed (Figure 8A–F): height 123 cm (<<3rd centile), weight 28 kg (<3rd centile), head circumference 52 cm (10th centile), span 113 cm, span/height ratio 0.91. Pubertal stage were A0P1B1, with testicular volume of 1-2 ml bilaterally. He had sparse hair, high and prominent forehead with high hairline, mild bitemporal depression, fairly large ears, blue-gray sclerae, malar hypoplasia, hypoplastic columella, short philtrum, small misaligned teeth in the lower jaw, flat palate with longitudinal median mucous thickening (Figure 8A–B). He had thin, pale, soft skin, with slightly prominent veins on trunk, limited elbow extension, skin hyperextensibility and distal joint hypermobility especially of the hands (Beighton scale: 6), long and tapered fingers with a tendency to ulnar deviation (Figure 8C–E), hypoplastic nails on the feet (Figure 8F).

**Figure 8.** Patient 3 **A**,**B**) Sparse hair, high and prominent forehead with high hairline, mild bi-temporal depression, fairly large ears, blue-gray sclerae, malar hypoplasia, hypoplastic columella, short philtrum, thin, pale, soft skin with prominent veins on the trunk; **C**–**E**) Skin hyperextensibility and distal joint hypermobility especially of the hands, long and tapered fingers, with a tendency to ulnar deviation; **F**) Feet: hypoplastic nails.

Skeletal X-rays demonstrated severe kyphoscoliosis, osteopenia (especially in the acetabula, femurs, tibiae and fibulae), and thin metacarpals, metatarsals and phalanges (Figure 9A–F).

Audiometric and ophthalmological evaluations were normal. Abdominal ultrasound was normal. Echocardiography demonstrated an atrial septal defect and mild mitralic insufficiency with thickened mitralic valve.


**Table 2.** Patients with *B3GALT6* variants: Review of the literature.


**Table 2.** *Cont.*

116

di

fficult data

interpretation;

 they include an additional variant, c.235A>C, p.(T79A).

#### *Genes* **2019**, *10*, 799

#### 3.3.2. Molecular Analysis

Karyotype was normal, 46XY. Genetic testing revealed three variants in the *B3GALT6* gene: NM\_080605:c.[308C>T;353delA];[987\_989delCTG], NP\_542172:p.[(Ala103Val);(Asp118Alafs\*160)]; [(\*330Alaext\*73)] (Figure 7A–C).

Variant c.353delA, already described as pathogenic (patient 2 and Nakajima et al., 2013 [45]), has been found in *cis* with variant c.308C>T, both inherited from the healthy father. This missense variant is absent from the reference population databases (gnomAD, 1000 genomes), but occurs at a poorly conserved position and is predicted as benign by multiple algorithms (MutationTaster, PolyPhen2, SIFT). It should be classified as a variant of uncertain significance, as there is not enough evidence to suggest whether this variant alone would be tolerated or damaging to protein function, or to indicate a possible contribution to the deleterious effect of the frameshift variant.

Variant c.987\_989delCTG, inherited from the healthy mother, disrupts the constitutive stop codon leading to an extended open reading frame, encoding a 72aa longer protein. It is absent from reference population databases (gnomAD, 1000 genomes), and by analogy with the frameshift variants in this gene, it is expected to produce an unstable or non-functional protein. Since it was found in trans with a recognized deleterious mutation, this variant can be considered as likely pathogenic.

#### **4. Discussion**

Biallelic variants in the *B4GALT7* and *B3GALT6* genes are responsible for conditions characterized by a combined skeletal and connective tissue phenotype. Their recent classification as spEDS [46] provides a valid assistance for their diagnosis, but the observation of new patients and the study of the clinical signs of those already reported in the literature can further improve our knowledge. In an effort to better define both the similarities and the peculiarities of these syndromes, we reviewed the notable features of our three patients and compared them to those of molecularly confirmed *B4GALT7*- and *B3GALT6*-related cases reported in the literature (Tables 1 and 2) [18,21–24,29–33,44,45,48,51–56]

All three of our patients display a remarkably short stature. Pt. 2 and 3 are more severely affected, in part because of their kyphoscoliosis, with a stature progressing from moderately short (or normal for Pt. 3) at birth to well below the 3rd percentile during childhood. Nearly all patients reported in the literature (65/66 evaluated cases) share this feature, confirming short stature as a main aspect of spEDS.

Joint hypermobility and soft, doughy and hyperextensible skin are some of the most striking characteristics in our patients. These qualities are particularly remarkable in the hands and, to a lesser extent, in the feet, and helped us restrict the diagnosis to an EDS-related condition. Since these features are shared by more than 90% of cases reported in the literature (63/66 and 62/68, respectively), in our opinion their combination should be considered a major diagnostic criterion for spEDS.

Craniofacial features for both *B4GALT7*- and *B3GALT6*-related conditions are quite variable among our patients and the reported individuals for whom either pictures or descriptive data could be evaluated, and do not seem suggestive of a facial gestalt. There are, however, some recurring features that may sometimes help to set the two conditions apart: i. proptosis is more frequent in spEDS-*B4GALT7* (32/33 vs 20/32); ii. midface retrusion has been frequently noted in both conditions, but in spEDS-*B3GALT6* it is often associated with a prominent forehead (31/38), sometimes emphasized by sparse hair (12/26) as in the case of our Pt. 2 and 3, while it would be more appropriate to state that spEDS-*B4GALT7* patients have an entirely flat face, usually short or round (31/33) and with a narrow mouth (31/33). The progeroid aspect of spEDS-*B4GALT7* described in Kresse's child [24] and previously pointed out by Hernandez [21–23] was absent in all other patients, suggesting that it is not strongly associated with mutations in *B4GALT7*. As current terminology indicates, a progeroid facial appearance is the subjective interpretation of a series of facial features which should rather be evaluated individually. Some features such as thin skin or excessive wrinkles may occasionally appear in the description of the cases we reviewed, but never in a combination suggestive of premature aging, and none of these were observed in our patient. Therefore, as already suggested by other authors, we endorse the necessity to remove the term "progeroid type" from this syndrome, because it could be misleading.

The radiological findings are undoubtedly the most significant evidence in distinguishing *B4GALT7*-related and *B3GALT6*-related spEDS: the latter includes more numerous and pronounced elements of skeletal dysplasia, and for this reason has often been classified as such in the past. Radioulnar synostosis has been observed recurrently and almost exclusively in spEDS-*B4GALT7* (19/32 cases, including our patient, vs 3 documented cases of spEDS-*B3GALT6* [53,56]. On the other hand, most of the individuals affected by *B3GALT6*-related spEDS, including our patients, display severe kyphoscoliosis, usually congenital or early onset and progressive (36/39), and several of the skeletal changes associated with SEMDJL1, such as platyspondyly, short iliac bones, elbow dislocation with misalignment of the long bones (33/36). Osteopenia has been noted in both conditions. It has been reported more frequently in *B3GALT6*-related cases (22/26), usually in conjunction with fractures and luxations, as in the case of our patients. In *B4GALT7*-related cases there are significantly fewer records of this feature, but in most reports its absence or presence cannot be verified. Therefore, the actual prevalence of osteopenia is unknown and should probably deserve more consideration when referring spEDS suspects for a radiological exam.

A consequence of skeletal involvement (bowing of limbs) is considered one of the main criteria for suspecting spEDS, and it is possibly more common in the spEDS-*B4GALT7* series (30/32 vs 16/20); in fact, our Pt. 2 and 3 do not have this phenotype. Joint contractures of the upper limbs have been noted for both conditions, though they are usually reported at the elbow in spEDS-*B4GALT7* and at the hands in spEDS-*B3GALT6*. In our experience, it was actually our *B3GALT6*-mutated patients who displayed limited elbow extension, but this is a non-specific clinical sign that is often found in skeletal dysplasia.

Other clinical features appear to be peculiar of *B3GALT6*-related spEDS. The most notable are talipes equinovarus (23/37 patients, including ours), possibly with hypoplastic nails (cfr EDS classification of 2017 [46] and our patients 2 and 3), and abnormalities of the dentition (13/17 patients, including ours), sometimes with yellow-brown discoloration of the teeth (also observed in our Pt. 2).

Motor delay has been noted in the majority of reports in which developmental milestones were evaluated (27/36), including our three patients. It is most likely related to the recurrent skeletal anomalies (bowing of limbs, scoliosis, hip dysplasia) as well as to the frequent reports of hypotonia, also observed in all of our patients (36/48 evaluated cases). In order to overcome the difficulties in sitting and walking, these children often need assistance in the form of rehabilitation with appropriate physiotherapeutic interventions. Depending on the severity of the musculoskeletal features, some individuals may be able to walk without help at a later age than their peers, while others can only manage walking with help.

Developmental delay appears to be present in only about half of the documented spEDS cases (26/49). In many instances, psychomotor development is only mildly delayed, possibly because of the confounding effect of minor behavioral abnormalities. Of our three patients, Pt. 1 IQ has been demonstrated to be average, while Pt. 2 and 3 require a support teacher during school only because of a diminished ability to concentrate, and actually have good comprehensive and expressive skills.

In summary, among all of these clinical features, the most relevant in order to suspect spEDS, because of their rarity in other syndromic conditions, are:

i. the extreme distal joint hypermobility and soft, hyperextensible skin, particularly of the hands;

ii. the radiological signs, which are the main indicator for discriminating spEDS-*B4GALT7*, associated with radioulnar synostosis, and spEDS-*B3GALT6*, characterized by kyphoscoliosis (congenital or early onset and progressive) and by the skeletal signs of SEMDJL1 (platyspondyly, short iliac bones, elbow dislocation).

Current EDS classification includes a third type of spEDS, caused by biallelic defects in SLC39A13 (OMIM \*608735), which encodes a zinc transporter involved in connective tissue development [57]. (Radiological findings suggest a minor skeletal involvement compared to other spEDS, but overall,)

the small number of reported families is inadequate to discuss a differential diagnosis with good confidence [58,59].

The differential diagnosis of spEDS-*B3GALT6* may be placed with kyphoscoliotic EDS (kEDS), primarily owing to early onset kyphoscoliosis, but the two conditions are very different from the clinical point of view, because the latter is not characterized by significantly short stature, and the joint hypermobility with dislocations/subluxations involves mainly shoulders, hips and knees rather than hands and feet.

The differential diagnosis may be placed also with musculocontractural EDS (mcEDS), but the latter is characterized by congenital multiple contractures, characteristically adduction-flexion contractures, and typical craniofacial features, which are distinct from spEDS.

The main differential diagnosis of spEDS-*B4GALT7* can be given with arthrochalasia EDS (aEDS), which is characterized by congenital bilateral hip dislocation and laxity of small joints, but generally aEDS also presents subluxation of the knees and dorso-lumbar kyphosis.

A clinician should consider the differential diagnosis between spEDS-*B4GAL7* and the *B3GAT3*-related linkeropathy, mainly because of prominent eyes, short stature with joint laxity and radioulnar synostosis. However, it should be noted that patients with *B3GAT3* variants often have cardiovascular abnormalities, and that the rarity of the cases described in the literature does not allow definite conclusions [14,15]. Additional clinical descriptions of these disorders are required in order to characterize the linkeropathies both individually and as a disease group.

The radiological feature of spEDS may further warrant consideration for differential diagnosis with several other skeletal dysplasias. The main discriminating features of *B4GALT7*- and *B3GALT6*-related conditions, i.e. respectively radioulnar synostosis and scoliosis with early onset and rapid evolution, can be found in various skeletal dysplasias, but are associated with such a remarkable degree of joint hypermobility and soft, hyperextensible skin almost exclusively in spEDS.

#### **5. Conclusions**

The most striking aspect of spEDS is the combination of clinical signs affecting both the connective tissue and the skeletal system. The criteria identified in the International Classification of EDS by Malfait et al. (2017) clearly show that there is significant overlap between the clinical features of spEDS-*B4GALT7* and spEDS-*B3GALT6*. This is confirmed and expanded upon by the clinical reports and literature review tables (Tables 1 and 2) presented here.

Overall, although skin and joints are similarly affected in both conditions, *B3GALT6* mutations lead to a more extensive and severe involvement of the skeletal system, with features often found in SEMDJL1 such as kyphoscoliosis, platyspondyly, short iliac bones and elbow disclocation.

Careful observation of the hands, with their very soft and hyperextensible skin and extreme distal joint laxity, in combination with the specific radiological signs, can immediately evoke the suspicion of either spEDS-*B4GALT7* (radioulnar synostosis) or spEDS-*B3GALT6* (severe progressive kyphoscoliosis and other SEMDJL1-like skeletal features).

The extreme hypermobility of distal joints and the soft, doughy skin on the hands and feet are rarely seen in other EDS types (except, to some degree, in kEDS and aEDS) and are a valuable clue for the diagnosis, which should be supported by skeletal radiographs and by molecular analysis.

Once spEDS is suspected, direct Sanger sequencing is still a viable and cost-effective option for molecular analysis, since *B4GALT7* and *B3GALT6* have short coding regions and only a few exons (6 and 1, respectively). This should be considered especially if the radiological signs restrict the hypothesis to either gene. However, particularly when attempting a molecular analysis early on, when some of the distinctive clinical signs have not yet evolved, an NGS panel specific for EDS would be advisable, since it can help confirm the differential diagnosis.

Being able to distinguish between spEDS-*B3GALT6* and spEDS-*B4GALT7* is important for clinicians, because some patients in the *B3GALT6*-related group may develop life-threatening complications such as aortic dilatation, aneurysms and cervical spine instability.

In conclusion, accurate diagnosis will help in excluding other causes of peripheral hypotonia, such as neuromuscular disorders, and allow for appropriate physiotherapeutic interventions.

**Author Contributions:** L.G. and A.S.-F conceived the study. L.G., I.M., I.I., M.P. M.D.R., C.M. and S.G. performed the clinical diagnosis of the patient, genetic counselling and follow-up; M.V. and A.R. performed biochemical studies in skin fibroblasts; S.S. and S.F. investigated bone health parameters; B.C-X. and S.U. carried out the molecular analyses and interpreted the results; I.I. and S.G.C. researched the literature; L.G, I.I., and S.G.C. prepared the manuscript; A.S.F. edited and coordinated the manuscript. All authors discussed, read, and approved the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors also wish to thank the patients' family members for their cooperation in providing the medical data and photographs necessary for this publication, as well as the photographers Marco Bonazzi and Luca Valcavi. The authors are grateful for the contribution made by the Fondazione Cassa di Risparmio Manodori of Reggio Emilia.

**Conflicts of Interest:** All authors declare that there are no conflicts of interest concerning this work.

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

### **Further Defining the Phenotypic Spectrum of** *B3GAT3* **Mutations and Literature Review on Linkeropathy Syndromes**

### **Marco Ritelli 1, Valeria Cinquina 1, Edoardo Giacopuzzi 2, Marina Venturini 3, Nicola Chiarelli <sup>1</sup> and Marina Colombi 1,\***


Received: 25 June 2019; Accepted: 19 August 2019; Published: 21 August 2019

**Abstract:** The term linkeropathies (LKs) refers to a group of rare heritable connective tissue disorders, characterized by a variable degree of short stature, skeletal dysplasia, joint laxity, cutaneous anomalies, dysmorphism, heart malformation, and developmental delay. The LK genes encode for enzymes that add glycosaminoglycan chains onto proteoglycans via a common tetrasaccharide linker region. Biallelic variants in *XYLT1* and *XYLT2*, encoding xylosyltransferases, are associated with Desbuquois dysplasia type 2 and spondylo-ocular syndrome, respectively. Defects in *B4GALT7* and *B3GALT6*, encoding galactosyltransferases, lead to spondylodysplastic Ehlers-Danlos syndrome (spEDS). Mutations in *B3GAT3*, encoding a glucuronyltransferase, were described in 25 patients from 12 families with variable phenotypes resembling Larsen, Antley-Bixler, Shprintzen-Goldberg, and Geroderma osteodysplastica syndromes. Herein, we report on a 13-year-old girl with a clinical presentation suggestive of spEDS, according to the 2017 EDS nosology, in whom compound heterozygosity for two *B3GAT3* likely pathogenic variants was identified. We review the spectrum of *B3GAT3*-related disorders and provide a comparison of all LK patients reported up to now, highlighting that LKs are a phenotypic continuum bridging EDS and skeletal disorders, hence offering future nosologic perspectives.

**Keywords:** linkeropathies; *B3GAT3*; Larsen-like syndrome; *B4GALT7*; *B3GALT6*; spondylodysplastic Ehlers-Danlos syndrome; *XYLT1*; *XYLT2*; Desbuquois dysplasia; spondylo-ocular syndrome

#### **1. Introduction**

Ehlers-Danlos syndrome (EDS) comprises a clinically variable and genetically heterogeneous group of heritable connective tissue disorders (HTCDs) sharing the triad of (generalized) joint hypermobility, cutaneous abnormalities, and internal organ/vascular fragility and dysfunctions. The 2017 EDS nosology recognizes 13 different clinical subtypes and 19 causal genes mainly encoding fibrillar collagens, collagen-modifying proteins, or processing enzymes [1]. A fourteenth subtype has been recently associated with biallelic variants in *AEBP1*, which encodes the aortic carboxypeptidase-like protein (ACLP) associating with collagens in the extracellular matrix (ECM) [2–5]. In addition to the clinical classification, the 2017 EDS nosology introduced a pathogenetic scheme that regrouped the EDS subtypes into seven functional classes (i.e., disorders) of (a) collagen primary structure and collagen processing, (b) collagen folding and crosslinking, (c) structure and function of the myomatrix, (d) glycosaminoglycan (GAG) biosynthesis, (e) intracellular pathways, (f) the complement pathway, and (g) unresolved forms of EDS. The clinical manifestations of EDS are broad and often overlap

other HCTDs including some types of skeletal dysplasias, cutis laxa, hereditary myopathies, and TGFβ-related disorders [6,7]. Hence, intermediate or bridging phenotypes presenting the EDS triad are expected at the boundaries of an evolving nosology.

The multisystemic clinical variability of EDS reflects the numerous functions of collagens and their interactors, among which proteoglycans (PGs) are particularly notable. PGs are structurally complex biomacromolecules that are essential in the development, signaling, and homeostasis of many tissues and organs including bone, cartilage, skeletal muscle, eyes, heart, and skin [8–13]. PGs contain one or more variable GAG chains, which are linear polysaccharides consisting of repeating disaccharide blocks attached to a core protein. Depending on the composition of these blocks, the PG superfamily can be subdivided into two major groups: heparan sulfate (HS) and chondroitin sulfate (CS)/dermatan sulfate (DS) PGs. The biosynthesis of GAG chains starts with the formation of a common so-called tetrasaccharide linker region that is covalently attached to a serine residue of the PG core protein (Figure 1). The linker region synthesis is a stepwise process that involves the action of specific glycosyltransferases. It starts with the transfer of a xylose (Xyl) residue by xylosyltransferases I/II (XylT-I/II encoded by *XYLT1* and *XYLT2*, respectively), followed by the addition of two galactose (Gal) residues by galactosyltransferase type I (GalT-I encoded by *B4GALT7*) and type II (GalT-II encoded by *B3GALT6*). The linker region is completed by the transfer of glucuronic acid (GlcA) catalyzed by glucuronosyltransferase I (GlcAT-I encoded by *B3GAT3*), upon which polymerization of the HS or CS/DS chains begins. HS is formed by the alternating addition of disaccharides of N-acetyl-glucosamine (GlcNAc) and GlcA residues, CS by N-acetyl-galactosamine (GalNAc) and GlcA residues, subsequently modified by several sulfotransferases. The formation of DS requires the epimerization of GlcA toward iduronic acid (IdoA), an event accomplished by dermatan sulfate epimerases (DS-epi1 encoded by *DSE*). This allows dermatan 4-o-sulfotransferase 1 (D4ST1 encoded by *CHST14*) to catalyze the 4-o-sulfation of GalNAc, which prevents back-epimerization of the adjacent IdoA (Figure 1) [8–13].

The importance of the correct initiation of GAG synthesis is exemplified by the identification of biallelic variants in all genes encoding the key enzymes in the linker region synthesis, leading to a spectrum of severe multisystemic disorders (Figure 1), a.k.a., linkeropathies (LKs) [11–13].

Two of these LKs, GalT-I- and GalT-II-deficiency, fit with the EDS spectrum and are recognized in the 2017 EDS nosology as spondylodysplastic EDS (spEDS) that also includes patients with mutations in *SLC39A13* encoding the ZIP13 protein involved in the influx of zinc into the cytosol [1,14]. According to the 2017 nosology, spEDS is suggested by two major criteria, short stature and muscle hypotonia, plus characteristic radiographic abnormalities and at least three other minor criteria (general or gene-specific) [1]. At present, 10 patients with molecularly confirmed *B4GALT7*-spEDS have been reported [14–21], 46 with *B3GALT6*-spEDS [14,22–28], and nine with *SLC39A13*-spEDS [14,29–31]. A further 22 patients, all with the same homozygous *B4GALT7* p.(Arg270Cys) missense variant, have been characterized in the ethnic group called white creoles living on Reunion Island (Larsen of Reunion Island syndrome) [32].

Molecular defects in *XYLT1* have been associated with Desbuquois dysplasia type 2 (DBQD2) with 28 molecularly proven patients reported hitherto [33–40], while *XYLT2* mutations cause the so-called spondylo-ocular syndrome (SOS), which has been described in 20 patients thus far [41–46]. Concerning *B3GAT3*, 25 patients from 12 families have been recognized, but range in severity from mild to severe and resemble Larsen (LR)-, Antley-Bixler (AB)-, Shprintzen-Goldberg (SG)-, and Geroderma osteodysplastica (GO)-like syndromes [26,47–53]. Further downstream in the GAG biosynthetic pathway, mutations in the *CHST14* and *DSE* genes have been reported in musculocontractural EDS (mcEDS) type 1 and 2, respectively [14].

Herein, we report on a 13-year-old girl ascertained as a suspect of EDS. Trio-based exome sequencing (ES) revealed compound heterozygosity for two likely pathogenic variants in *B3GAT3*. We provide a comparison of the patient's clinical features with those of the other LK patients reported so far, thus offering future perspectives for clinical research in this field.

**Figure 1.** Biosynthetic assembly of the glycosaminoglycan (GAG) backbones of heparan sulfate (HS) and chondroitin sulfate (CS)/dermatan sulfate (DS) chains and related genetic disorders. Each enzyme (in red) and its coding gene (in black) are described near the sugar symbols. After the synthesis of specific core proteins, the synthesis of the GAG-protein linker region is initiated by XylT-I/II, which transfers [Xyl] to the specific Ser residue in the endoplasmic reticulum. The synthesis of the linker region is completed by the consecutive addition of two molecules of [Gal], added by GalT-I/II, followed by the transfer of [GlcA] catalyzed by GlcAT-I in the Golgi. The addition of a [GalNAc] to the linker region commits the growing GAG chain to CS/DS. CS synthesis proceeds with the alternating addition of [GlcA] and [GalNAc] and can be further modified by sulfotransferases. A DS chain is generated after the formation of the chondroitin backbone, when [GlcA] is converted into [IdoA] by DS-epi1, resulting in the formation of the dermatan backbone, where [GalNAc] is sulfated by D4ST1. Alternatively, the addition of the [GlcNAc] to the linker region induces HS biosynthesis. The polymerization of the HS chain is catalyzed by enzymes encoded by the *EXT1* and *EXT2* genes. **Abbreviations**: Ser, Serine; HS, heparin sulfate; CS, chondroitin sulfate; DS, dermatan sulfate; XylT-I/II, xylosyltransferases I/II; GalT-I, galactosyltrasferase I; Galt-II, galactosyltrasferase II; GlcAT-I, glucuronasyltransferase I; DE-epi1; dermatan sulfate epimerases; D4ST1; dermatan 4-o-sulfotransferase 1; Xyl, xylose; Gal, galactose, GlcA; glucuronic acid, GlcNAc, *N*-acetylglucosamine; GalNAc, *N*-acetylgalactisamine; IdoA, iduronic acid; DBQD2, Desbuquois dysplasia type 2, SOS, spondylo-ocular syndrome; spEDS, spondylodysplastic EDS; LRS-like, Larsen-like syndrome; ABS-like, Antley-Bixler-like syndrome; SGS-like, Shprintzen-Goldberg-like syndrome; GO-like, Geroderma osteodysplastica-like; mcEDS, musculocontractural EDS.

#### **2. Patient and Methods**

#### *2.1. Ethical Compliance*

The patient was evaluated at the specialized outpatient clinic for the diagnosis of EDS and related connective tissue disorders (i.e., the Ehlers-Danlos Syndrome and Inherited Connective Tissue Disorders Clinic, CESED), at the University Hospital Spedali Civili of Brescia. Molecular analysis was achieved in compliance with Italian legislation on genetic diagnostic tests and the patient's parents provided written informed consent for the publication of clinical data and photographs according to the Italian bioethics laws. This study followed the Declaration of Helsinki's principles and was carried out from routine diagnostic activity; a formal ethics review was therefore not requested.

#### *2.2. Amplicon-Based Exome Sequencing*

Genomic DNA from the proband and her parents was extracted from peripheral blood leukocytes by standard procedures. Mutational screening was performed by trio-based ES using the Ion Proton platform and the AmpliSeq technology following the manufacturer's recommendations (Thermo Fisher Scientific, South San Francisco, CA, USA). Briefly, whole exome libraries were prepared using the AmpliSeq Exome RDY kit for library preparation. The template preparation of the libraries was performed using the Ion PI Hi-Q OT2 200 kit on the Ion OneTouch 2 starting from 8 μl of the 100 pM libraries. Template preparation and sequencing runs were performed with the Ion PI Hi-Q Sequencing 200 kit. The templated Ion Sphere Particles (ISP) were enriched for positive ISP using the Ion OneTouch ES and sequenced on the Ion Proton with the Ion PI chip v3. Basecalling and sequence alignment against hg19 genome assembly were performed using Ion Torrent Suite software 5.6, and genetic variants were identified using Torrent Suite Variant Caller pipeline 5.6. Variants were decomposed and normalized using the vt tool [54], filtered for quality using GARFIELD-NGS [55], and annotated using ANNOVAR [56]. Variants were filtered according to the following criteria: (i) MAF < 0.01 in 1000G and ExAc v.0.3 populations; (ii) predicted to alter protein product, namely missense, stop-affecting or splice-affecting variants; and (iii) not present in our internal database. Filtered variants were then prioritized based on DANN [57] and M-CAP [58] scores to retain the most likely deleterious variants. RVIS [59] and GDI [60] scores were used to prioritize more intolerant genes. Finally, we only considered variants with a perfect segregation among the parents and proband according to a recessive/de novo model of transmission. To evaluate the putative pathogenicity of the *B3GAT3* variants, we used the following mutation prediction programs: SIFT [61], Mutation Taster [62], CADD [63], PROVEAN [64], GERP++ [65], UMD\_prediction [66], LRT [67], Fathmm-MKL [68], VEST [69], and FitCons [70].

#### *2.3. Sanger Sequencing*

The *B3GAT3* variants identified by ES (reference sequences: NM\_012200.3, NP\_036332.2) were confirmed by Sanger sequencing with the BigDye Terminator v1.1 Cycle Sequencing kit on an ABI 3130XL Genetic Analyzer according to the manufacturer's protocols (Thermo Fisher Scientific, South San Francisco, CA, USA) with specific primer sets amplifying exon 3 and exon 4 of *B3GAT3*, respectively (Supplementary Table S1). The sequences were analyzed with Sequencer 5.0 software (Gene Codes Corporation, Ann Arbor, MI, USA) and variants were annotated according to the Human Genome Variation Society (HGVS) nomenclature by using the Alamut Visual software version 2.11 (Interactive Biosoftware, Rouen, France).

#### **3. Results**

#### *3.1. Clinical Findings*

The proband (LOVD ID #00235371), an Italian 13-year-old girl, was born to non-consanguineous healthy parents and had a healthy sister. Clinical history was remarkable for birth at 41 weeks (height 49 cm, weight 3.2 kg) after induced labor associated with perinatal respiratory distress, anterior ectopic anus, and congenital hip dislocation treated with hip abduction braces, severe neonatal hypotonia, and delayed motor development (delay in walking, first step at three years of age, and acquisition of fine motor skills). She was discharged from the neonatology unit with a diagnosis of generalized joint hypermobility. Medical history further included propensity to develop ecchymoses and surgically treated umbilical hernia. At one year of age, total skeletal x-ray disclosed severe kyphoscoliosis unsuccessfully treated with orthopedic corset, atlantooccipital instability, bilateral radio-ulnar synostosis, abnormalities of the proximal humeral epiphyses, metaphyseal flaring, long and thin bones with widened metaphyses, and delayed bone age. Cervical spine imaging showed significant atlantoaxial and atlanto-occipital instability with flexion and extension. At the age of two, a heart ultrasound revealed an atrial septal defect, which was surgically treated three years later due to tachyarrhythmia and pulmonary hypertension. At ages two and three, respectively, either a clinical

diagnosis of SGS or of an unspecified EDS was given. Genetic analyses were not performed. At age four, dual-energy x-ray absorptiometry (DEXA) disclosed severe low bone mineral density for sex and age (z-score < 2 SD) and bisphosphonate treatment was commenced. The progressive kyphoscoliosis and cervical spine instability were surgically treated at age six and 12 without satisfactory improvement. Ophthalmologic evaluation revealed refractive errors such as astigmatism and strabismus. At six years of age, progressive height deficit related to GH deficiency was noticed. GH therapy was started with a discrete outcome, but treatment was interrupted two years later due to the worsening of side effects. Since infancy, the patient suffered from recurrent dislocation of the elbows, shoulders, and knees as well as chronic myalgia of the lower limbs and severe foot pain.

On examination, at 13 years of age, she presented with short stature (height 130 cm; genetic target 169 cm, arm span/height ratio, normal value < 1.05), a weight of 25 kg, severe kyphoscoliosis, short neck, pectus carinatum, genua valga, and muscle hypotonia (Figure 2). Facial dysmorphism included enophthalmos, midface hypoplasia, dolichocephaly, prominent forehead, low-set ears, blue sclerae, downslanting palpebral fissures, long philtrum, narrow palate, and micrognathia. Foot deformities such as sandal gaps, severe pes planovalgus, and clinodactyly of the toes were present (Figure 2). The patient also showed skin hyperextensibility over the neck, elbow, forearm and knees, easy bruising, mild atrophic scarring, generalized joint hypermobility with a Beighton score (BS) of 6/9, clinodactyly of the fifth finger, and long fingers with spatulate distal phalanges. Cognitive development and mentation were normal. DEXA confirmed severe osteopenia and delayed bone age despite bisphosphonates treatment; no fractures were reported. Echocardiogram revealed minimal mitral, tricuspid, and aortic valves insufficiency.

**Figure 2.** Clinical findings of the patient. Facial dysmorphism: enophthalmos, midface hypoplasia, prominent forehead, micrognathia, low-set ears, and short neck (**a**), skin hyperextensibility over the neck (**b**), long fingers with spatulate distal phalanges and clinodactyly of the fifth finger (**c**), joint laxity of the fifth finger (**d**), foot deformities: sandal gaps, severe pes planovalgus and clinodactyly of the toes (**e**), elbow deformity with reduction in the range of motion (**f**), radioulnar synostosis (**g**), severe kyphoscoliosis (**h**–**l**), pectus carinatum (**h**), muscle hypotonia (**h**–**k**), and metaphyseal flaring (**m**).

Overall, the patient fulfilled the minimal criteria suggestive for spEDS according to the 2017 EDS nosology [1], since she presented two major (short stature, muscle hypotonia), four general minor (skin hyperextensibility, foot deformity, delayed motor development, osteopenia), and several gene-specific minor criteria either for *B4GALT7*-spEDS or *B3GALT6*-spEDS (craniofacial dysmorphism, characteristic radiographic findings including severe kyphoscoliosis, radioulnar synostosis, metaphyseal flaring, bilateral elbow deformities, peculiar fingers, and gJHM with recurrent dislocations) (Table 1, Supplementary Table S2). Nevertheless, since the patient also presented some peculiar features not previously reported in spEDS patients such as dolichocephaly, atrial septal defect, and anterior ectopic anus, we performed trio-based ES.


**Table 1.** Candidate genes after selecting variants.

#### *3.2. Molecular Findings*

Summary results of ES are reported in Supplementary Table S3 and Figure S1. After application of the filtering pipeline and prioritization of variants by considering only recessive or de novo variants with perfect segregation among trio members (Supplementary Table S4), six candidate genes were identified (Table 1). Among these, five genes with a de novo variant were excluded, since only *B3GAT3*, associated with multiple joint dislocations, short stature, and craniofacial dysmorphism with or without congenital heart defects (OMIM #245600), was consistent with the patient's phenotype.

In particular, the trio analysis revealed the paternally inherited c.481C>T transition in exon 3, leading to the substitution of a highly conserved and positively charged arginine residue with a larger and neutral tryptophan at position 161 [p.(Arg161Trp)] within the donor substrate binding subdomain of the protein [71,72], and the maternal c.889C>T variant in exon 4, which also resulted in the substitution of an arginine residue with a tryptophan [p.(Arg297Trp)], but within the acceptor substrate binding subdomain (Figures 3A and 4) [71,72].

Both variants are annotated in dbSNPs and have extremely low frequencies in population genomic databases. In particular, the paternal variant was observed in three individuals in GnomAD (rs765246909, 3/251290, no homozygotes, total MAF: C = 0.00001194), and the maternal substitution was also observed in three individuals (rs759636773, 3/251080, no homozygotes, total MAF: C = 0.00001195) (queried on 28 May, 2019). Their putative pathogenicity was estimated through 12 different in silico prediction algorithms that agreed to define p.(Arg161Trp) and p.(Arg297Trp) as high impacting variants (Figure 3B).

By using the InterVar (Clinical Interpretation of Genetic Variants) tool [73], both variants were classified as likely pathogenic (class 4) according to the guidelines of the American College of Medical Genetics and Genomics (ACMG) [74] since (i) both variants are missense substitutions in a gene that has a low rate of benign missense variation and where missense variants are a common mechanism of disease (Table 2); (ii) both variants are located in a critical and well-established functional domain of the protein, i.e., the glycosyltransferase domain; (iii) their extremely low frequency in a publicly available population database; (iv) the multiple lines of computational evidence supporting a deleterious effect; and (v) the patient's phenotype was highly suggestive for a disease with a single genetic etiology.

**Figure 3.** Molecular findings. (**A**) ES data alignments show the compound heterozygosity of the paternal c.481C>T (p.Arg161Trp) and the maternal c.889C>T (p.Arg297Trp) missense variants. Sanger sequencing confirmed the presence of both variants in the proband (arrows). Healthy parents were heterozygous carriers. Mutations are annotated according to HGVS nomenclature (reference sequences: NM\_012200.3, NP\_036332.2). (**B**) In silico predictions of the pathogenicity of the p.(Arg161Trp) and p.(Arg297Trp) missense substitutions by using 12 different algorithms [61–70].

The two variants were submitted to the Leiden Open Variation Database (LOVD variants identifiers: #0000480198 and #0000480199). The functional effect of the missense substitutions on reduced/absent enzymatic activity was not verified due to the unavailability of the patient's fibroblasts.


ofclinicalfeaturesofallwith

#### *Genes* **2019** , *10*, 631



*Genes* **2019** , *10*, 631


**Table 2.** *Cont*.

134

#### **4. Discussion**

The umbrella term LK refers to a group of extremely rare and consequently poorly characterized genetic disorders caused by mutations in genes encoding enzymes responsible for the synthesis of GAG side chains of PGs. Nosologic uncertainty characterizes these disorders, thus contributing to the clinical diagnosis of challenging patients, which is not straightforward at all. Indeed, although the linker region is the identical tetrasaccharide sequence for all PGs in all tissues, biallelic variants in the LK genes are associated with apparently different phenotypes that variably affect the skeletal system and skin, even if remarkable similarities between the different LKs are recognizable [8–13].

In the 2017 EDS nosology, some patients with defects in two out of the five LK genes (i.e., *B4GALT7* and *B3GALT6*) were grouped as spEDS together with those harboring *SLC39A13* mutations, in consideration of the reliable clinical overlap [1], whereas the 22 Larsen of Reunion Island syndrome patients were not included. We have previously suggested that, though some phenotypic variations between Larsen of Reunion Island syndrome and *B4GALT7*-spEDS exist, these conditions should not be considered as different entities [20]. Patients with *B3GAT3*, *XYLT1*, and *XYLT2* mutations were also not classified as spEDS, even though there is a common pathogenic mechanism and numerous shared clinical features. The patient reported in this study corroborates the awareness that LKs are a phenotypic continuum bridging EDS and skeletal disorders. Indeed, the patient was referred to our clinic with a well-founded suspicion of EDS, since she respected the EDS triad and fulfilled the minimal suggestive criteria of spEDS (Supplementary Table S2). In consideration of some peculiar signs not previously associated with spEDS, we performed ES, which revealed compound heterozygosity for two likely pathogenic *B3GAT3* variants.

*B3GAT3* is involved in a spectrum of connective tissue and skeletal disorders. Table 2 summarizes the clinical features of the 26 patients from 13 different families, 11 of which were consanguineous with *B3GAT3* mutations reported so far ([26,47–53], present study). Among the *B3GAT3*-related disorders, a LRS-like presentation similar to that of our patient was the most common, but more severe phenotypes resembling ABS, SGS, and GO have also been reported (Table 2).

Historically, Baasanjav et al. [47] first described five patients with short stature, radioulnar synostosis, brachycephaly, and cardiac abnormalities. The authors suggested naming this condition as Larsen-like syndrome, *B3GAT3* type. All patients carried the homozygous c.830G>A, p.(Arg277Gln) missense mutation in the acceptor substrate binding subdomain of the protein. Von Oettingen et al. [48] described a 5-year-old boy with a similar phenotype and the same missense variant. Novel findings were developmental delay, refractive errors, pectus carinatum, atlantoaxial and atlanto-occipital instability, and excessive skin wrinkling. Budde et al. [49] reported eight patients from a large consanguineous family with a LRS-like phenotype without cardiac involvement carrying the c.419C>T, p.(Pro140Leu) pathogenic variant in the donor substrate binding subdomain of the protein. Job et al. [51] described the first compound heterozygous patient, a 6-year-old boy, who presented in addition to the typical LRS-like features, hypotonia, hyperextensible skin, and generalized osteoporosis with multiple fractures. The identified mutations were a null allele (c.1A>G, p.Met1?) and the c.671T>A, p.(Leu224Gln) missense substitution in the acceptor substrate binding subdomain of the protein, respectively. Very recently, Colman et al. [53] characterized a 13-year-old girl with a rather mild phenotype with short stature, short neck, craniofacial dysmorphism, joint hypermobility with dislocation, foot deformities, and mild osteopenia without fractures. The patient was homozygous for the c.416C>T, p.(Thr139Met) missense variant in the donor substrate binding subdomain.

Alazami et al. [26] reported a GO–like syndrome in a patient carrying the homozygous c.245C>T, p.(Pro82Leu) missense variant in the donor substrate binding subdomain of the protein. A detailed clinical description is lacking, but short stature, spondyloepimetaphyseal dysplasia, cutis laxa, generalized osteoporosis with fractures, and several bony chondromas were reported.

Finally, Jones [50], Yauy [52], and Colman et al. [53] described the most severely affected patients reported hitherto, all harboring the same c.667G>A, p.(Gly223Ser) missense mutation in the acceptor substrate binding subdomain of the protein. Jones et al. [50] reported a 12-month-old boy with short stature, hypotonia, global developmental delay, radioulnar synostosis, metaphyseal flaring, craniofacial dysmorphism, sandal gap, bilateral club feet, septal defects, and multiple fractures. Novel findings included blue sclerae, bilateral glaucoma, diaphragmatic hernia, small chest, arachnodactyly, lymphedema, hearing loss, and perinatal cerebral infarction. Colman et al. [53] characterized an infant, who died at the age of 2.5 months and showed cutis laxa, contractures of large and small joints, finger and foot anomalies, short neck, severe asymmetric thorax, dolichocephaly and other facial dysmorphic features, multiple long bones fractures, and bilateral corneal clouding. Likewise, Yauy et al. [52] reported six patients, who all died before one year of age, with craniosynostosis, midface hypoplasia, radioulnar synostosis, multiple neonatal fractures, dislocated joints, joint contractures, and cardiovascular abnormalities. The authors suggested that a *B3GAT3*-related disorder with craniosynostosis and bone fragility should be considered as a differential diagnosis in the prenatal period for ABS and in the postnatal period for SGS. Indeed, ABS is suspected during pregnancy if ultrasonography shows craniofacial deformities due to craniosynostosis (a hallmark of ABS), midface hypoplasia, bilateral radiohumeral, or radioulnar synostosis [52,75]. In the postnatal period, SGS can also be suspected, based on the association of craniosynostosis and arachnodactyly. Indeed, SGS is characterized by craniosynostosis and distinctive craniofacial features, skeletal anomalies (arachnodactyly, dolichostenomelia, camptodactyly, pes planus, pectus excavatum or carinatum, kyphoscoliosis), cardiovascular anomalies, intellectual deficiency, and brain anomalies [76].

Interestingly, our patient received a diagnosis of SGS in infancy, though there was the absence of several cardinal features, above all intellectual disability and brain abnormalities. Overall, the clinical presentation of our patient is consistent with a moderate-severe LRS-like phenotype, characterized by short stature, hypotonia, joint hypermobility with recurrent dislocations, radioulnar synostosis, peculiar fingers, foot deformities, midface hypoplasia, short neck, and cardiovascular abnormalities. Of note, although severe osteopenia was present, our patient never experienced fractures. Other features observed in our patient such as skin hyperextensibility, pectus carinatum, atlantoaxial and atlanto-occipital instability, severe kyphoscoliosis, and blue sclerae have been rarely reported in other individuals with *B3GAT3* mutations, while easy bruising, atrophic scarring, and anterior ectopic anus are novel findings (Table 2). Since most of the initial reports focused on a particular aspect of the phenotype, mostly skeletal features, it remains possible that cutaneous and other systemic features that were present have not been described.

The heterogeneous LK phenotypes, ranging from mild to severe and even lethal presentations, seem to be related to specific *B3GAT3* mutations, though there has been a limited number of patients and pathogenic variants described so far (Table 2, Figure 4) [52,53].

In particular, the more severe phenotypes appear to harbor mutations located within the acceptor substrate binding subdomain of the catalytic domain of the protein, whereas more mildly affected patients seem to have mutations in the donor substrate binding subdomain [52,53,77]. The intermediated phenotype of our patient, compound heterozygous for missense variants in both these subdomains (p.(Arg161Trp) in the donor and p.(Arg297Trp) in the acceptor substrate binding subdomain, respectively), corroborates this preliminary genotype–phenotype correlation.

Concerning the p.(Arg161Trp) missense variant, Fondeur-Gelinotte et al. [78] previously demonstrated the importance of this residue in the uridine-diphosphate (UDP)-GlcA donor binding site by site-directed mutagenesis and biochemical analyses. Concerning the consequence of the p.(Arg297Trp) missense variant in the acceptor substrate binding subdomain, in silico modeling using the Hope software [79] suggested that its 3D structure should be severely compromised. Indeed, the wild type residue forms either a hydrogen bond with a glutamic acid at position 295 or a salt bridge with glutamic acid at positions 206 and 295. The differences in size and hydrophobicity between the wild type and mutant residue are predicted to perturb both the hydrogen bond and the ionic interaction, thus likely abolishing the GlcAT-I activity.

**Figure 4.** Schematic illustration of the *B3GAT3* structure and glucuronosyltransferase I protein domains. The different *B3GAT3* pathogenic variants found in all patients reported thus far [1-8, present study] are shown over the diagram. The variants identified in this study are in red. Variants are annotated according to HGVS nomenclature (reference sequences: NM\_012200.3, NP\_036332.2). (**Below**) The protein structure of GlcAT-I is reported with its multiple domains. In different grey scales from the N-terminus (NH2) to the C-terminus (COOH), the small cytoplasmic domain (residues 1–7), the transmembrane domain (residues 8–25), and the proline-rich stem region (residues 26–74); in blue and green the catalytic domain consisting of the UDP-GlcUA (uridine diphosphate—(β-D-) glucuronic acid) donor substrate binding subdomain (residues 75–197, in blue), and the acceptor substrate binding subdomain (residues 198–308, in green), according to references [71,72,78].

The impression that LKs, even though likely not fully characterized yet, should be considered as a phenotypic continuum emerges from the comparison of all LK patients reported up to now (Table 3), which shows several overlapping clinical features, but also some differences. These data support the idea that the enzymes involved in the biosynthesis of the PG linker region may be part of a larger enzyme complex rather than functioning independently, and that the existing phenotypic disparity could be due to varying levels and spatiotemporal gene expression in different tissues [23,48].


**Table 3.** Clinical features of linkeropathies.

**Table 3.** *Cont*.


Note: na, not available; +, present in all patients; -, absent in all patients; a: Peculiar fingers including long, slender, tapered, broad, thin, arachnodactyly; b: Foot deformity including pes planus, hallux valgus, club feet, sandal gap; *B3GAT3*: ([26,47–53], present patient); *B4GALT7*: [15–21,32]; *B3GALT6*: [22–28], *SLC39A13*: [29–31], *XYLT1*: [33–40]; *XYLT2*: [41–46].

The phenotypic features with a high rate of incidence shared among all LKs include short stature, joint laxity with dislocations, craniofacial dysmorphism (especially prominent forehead/eyes and blue sclerae), pectus abnormalities, peculiar fingers, foot deformities, and to a variable degree hypotonia and developmental delay. Delayed cognitive development is instead more frequently observed in *XYLT1*-related DBQD2, followed by *XYLT2*-related SOS, and *B4GALT7*- and *B3GALT6*-spEDS (Table 3).

While joint hypermobility is a common trait of all LKs, joint contractures are more frequently observed in patients with *B3GAT3* and *B3GALT6* mutations. Low bone mineral density/osteopenia and radiographic abnormalities are also very common, whereas generalized osteoporosis and multiple fractures are more frequently encountered in *B3GALT6*-spEDS and in SOS. Likewise, among the numerous radiographic findings, severe progressive kyphoscoliosis is particularly frequent in *B3GALT6*-spEDS, SOS, and DBQD2. In addition to progressive (kypho)scoliosis, short long bones with a monkey wrench or Swedish key appearance of the femora, and advanced carpal and tarsal ossification are also more frequently observed in DBQD2 (Table 3). Of note, the presence or absence of specific hand anomalies, comprising an extra-ossification center distal to the second metacarpal, delta phalanx, or bifid distal thumb phalanx, together with dislocations of the interphalangeal joints are considered radiographic hallmarks distinguishing DBDQ1 (resulting from pathogenic variants in *CANT1)* from DBDQ2 [34]. Platyspondyly, a hallmark of "spondylo"-dysplasia, has never been recognized in *B4GALT7*-spEDS and *B3GAT3*-related disorders, except in the patient reported by Alazami et al. [26], whereas it is frequently reported either in *B3GALT6*- and *SLC39A13*-spEDS or in SOS, whereas in DBDQ2 it was recognized in five out of the nine investigated patients. Radioulnar synostosis seems more specific for *B3GAT3*-and *B4GALT7*-related disorders, the latter including either *B4GALT7*-spEDS (8/10) or Larsen of Reunion Island syndrome (10/21). Radiographic anomalies that are more common in *B3GALT6*-spEDS when compared to other LKs comprise metaphyseal flaring, iliac abnormalities, radial head subluxation/dislocation, which is also frequent in *B4GALT7*-spEDS, and bowing of limbs, which is shared with *SLC39A13*-spEDS (Table 3).

Among the plethora of craniofacial dysmorphism, midface hypoplasia, craniosynostosis, and short/webbed neck are particularly frequent in *B3GAT3*-related disorders. Wide forehead, hypertelorism, and microstomia are more common in *B4GALT7*-spEDS, whereas frontal bossing, micrognathia, and abnormal dentition characterize *B3GALT6*- and *SLC39A13*-spEDS. Flat face is shared among all spEDS subtypes and DBDQ2 (Table 3).

Ocular involvement might facilitate the differential between SOS and other LKs. In particular, the presence of cataract, retinal detachment has been described thus far only in patients with *XYLT2* mutations. Refractive errors/hypermetropia is also frequent in SOS, but has also been found in *B4GALT7*-spEDS and *B3GAT3*-disorders. Apart from eye involvement, deafness is also more frequent in SOS when compared to the other LKs (Table 3).

Marked cutaneous anomalies including hyperextensible, soft, doughy, thin and translucent skin, and atrophic scarring seem to distinguish spEDS patients, party justifying their inclusion within the EDS spectrum. On the other hand, in all the other LKs, several patients (including ours) with skin hyperextensibility and other abnormalities (including even cutis laxa-like features) have been published (Table 3). As above-mentioned for *B3GAT3*, it remains possible that in more than a few patients, the cutaneous involvement was not investigated. Therefore, further reports are needed to define the real incidence of skin abnormalities in LKs.

Concerning cardiovascular involvement, anomalies such as septal defects, aortic valve dysplasia, aortic root and ascending aorta dilatation, and mitral valve prolapse, are more recurrent in *B3GAT3*-related disorders as well as in SOS (Table 3).

#### **5. Conclusions**

In summary, our findings expand the *B3GAT3* allelic repertoire, corroborate the emergent genotype-phenotype correlations, and confirm the extended phenotypic range of *B3GAT3* mutations overlapping skeletal dysplasia and soft HCTDs including EDS. Furthermore, we provided a comprehensive overview of the phenotypic features of *B3GAT3*-related disorders and of all the LKs, thus offering future nosologic perspectives for either EDS or skeletal dysplasias. Given the convergent pathogenic pathway and the important clinical overlap not only among *B3GAT3*-associated diseases and spEDS (a term that, in our opinion, is unfortunate, since platyspondyly is not present in all subtypes), but also with DBDQ2 and SOS, this group of HCTDs should be considered as a phenotypic continuum and not as distinct entities.. In the genetic era of the classification of disorders, we propose that the term LK is preceded by the specific causal gene. Further reports on additional patients are awaited as well as functional studies on the spatiotemporal expression of the different glycosyltransferases to unravel the molecular mechanisms involved in the pathophysiology of LKs needed to identify potential therapeutic options.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/9/631/s1, Table S1. Primers. Table S2. Clinical features of the present patient and comparison with spEDS and *B4GALT7*-LRS. Table S3. Summary of exome sequencing results. Figure S1. Coverage distribution. Table S4. Variants with perfect segregation among trio members.

**Author Contributions:** M.C. and M.R. conceived the study. M.V. and M.C. performed the clinical evaluation of the patient, genetic counselling, and follow-up; M.R., V.C., and E.G. carried out the molecular analyses; M.R., V.C., and N.C. researched the literature; M.R., V.C., and N.C. prepared the manuscript; M.C. edited and coordinated the manuscript. All authors discussed, read, and approved the manuscript.

**Funding:** No funding was received for this project.

**Acknowledgments:** The authors wish to thank the patient for her cooperation during the diagnostic process and the Fazzo Cusan family for the generous support.

**Conflicts of Interest:** All authors declare that there are no conflicts of interest concerning this work.

#### **References**


Hart, G.W., Etzler, M.E., Eds.; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, USA, 2009; Chapter 16.


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **The Connective Tissue Disorder Associated with Recessive Variants in the** *SLC39A13* **Zinc Transporter Gene (Spondylo-Dysplastic Ehlers–Danlos Syndrome Type 3): Insights from Four Novel Patients and Follow-Up on Two Original Cases**

**Camille Kumps 1, Belinda Campos-Xavier 1, Yvonne Hilhorst-Hofstee 2, Carlo Marcelis 3, Marius Kraenzlin 4, Nicole Fleischer 5, Sheila Unger <sup>1</sup> and Andrea Superti-Furga 1,\***


Received: 2 April 2020; Accepted: 10 April 2020; Published: 14 April 2020

**Abstract:** Recessive loss-of-function variants in *SLC39A13,* a putative zinc transporter gene, were first associated with a connective tissue disorder that is now called "Ehlers–Danlos syndrome, spondylodysplastic form type 3" (SCD-EDS, OMIM 612350) in 2008. Nine individuals have been described. We describe here four additional affected individuals from three consanguineous families and the follow up of two of the original cases. In our series, cardinal findings included thin and finely wrinkled skin of the hands and feet, characteristic facial features with downslanting palpebral fissures, mild hypertelorism, prominent eyes with a paucity of periorbital fat, blueish sclerae, microdontia, or oligodontia, and—in contrast to most types of Ehlers–Danlos syndrome—significant short stature of childhood onset. Mild radiographic changes were observed, among which platyspondyly is a useful diagnostic feature. Two of our patients developed severe keratoconus, and two suffered from cerebrovascular accidents in their twenties, suggesting that there may be a vascular component to this condition. All patients tested had a significantly reduced ratio of the two collagen-derived crosslink derivates, pyridinoline-to-deoxypyridinoline, in urine, suggesting that this simple test is diagnostically useful. Additionally, analysis of the facial features of affected individuals by DeepGestalt technology confirmed their specificity and may be sufficient to suggest the diagnosis directly. Given that the clinical presentation in childhood consists mainly of short stature and characteristic facial features, the differential diagnosis is not necessarily that of a connective tissue disorder and therefore, we propose that *SLC39A13* is included in gene panels designed to address dysmorphism and short stature. This approach may result in more efficient diagnosis.

**Keywords:** Ehlers–Danlos syndrome; SLC39A13; dysmorphology; short stature; connective tissue; DeepGestalt technology

#### **1. Introduction**

The Ehlers–Danlos syndrome (EDS) is a group of disorders that affect connective tissues in the skin, ligaments, joints, blood vessels, and other organs. Defects in connective tissue cause a wide range of clinical manifestations, from mildly loose joints to life-threatening conditions, such as arterial bleeds. From the original description as a disorder of hyperelastic skin and lax joints at the beginning of the 20th century [1], a first classification with four main types was proposed in 1970 [2]; molecular advances have allowed the recognition of many distinct disorders that, although different from the classic EDS types described by Beighton, have been given the moniker of "Ehlers–Danlos syndrome" as a reflection of the presence of connective tissue fragility. Thus, the recent version of EDS classification has been expanded to include a wide range of disorders, including skeletal dysplasia [3]. However, the clinical criteria remain relatively non-specific, and clinical diagnosis can be difficult.

A connective tissue disorder associated with recessive biallelic variants in *SLC39A13*, and a mouse knock-out model for the same gene, were described in 2008 by two separate but collaborating groups [4,5]. The features noted in these eight patients were postnatal-onset short stature, protuberant eyes with bluish sclerae and down-slanting palpebral fissures, thin and moderately hyperelastic skin with bruisability, and hands with finely wrinkled palms, tapering fingers, thenar atrophy, and moderate hypermobility of the small joints. Skeletal radiographs showed a moderate degree of platyspondyly with irregular vertebral end plates as well as minor epimetaphyseal changes in appendicular bones. A reduced molar ratio of pyridinoline-to-deoxypyridinoline in urine was observed, indicating reduced collagen lysyl hydroxylation. The latter finding is typically observed in collagen lysyl hydroxylase deficiency (EDS type VI-A), but in that condition, there is severe muscular hypotonia from birth, progressive kyphoscoliosis, and normal height (apart from kyphoscoliosis). Additionally, there were no pathogenic mutations in *PLOD1*; instead, the patients reported by Giunta et al. (2008) were homozygous for an in-frame 9-bp deletion in *SLC39A13*, c.483\_491del (p.F162\_164del), while those reported by Fukada et al. (2008) were homozygous for the *SLC39A13* variant c.221G>A (p.G74D). The novel condition was given the name of "spondylo-cheiro-dysplastic EDS" for the distinguishing features of the hand and the platyspondyly [4,5]. Later, the EDS nosology has used the name "spondylodysplastic EDS" for a group of three conditions, B3GALT6 deficiency (better known as spondyloepimetaphyseal dysplasia with joint laxity, Beighton type), B4GALT7 deficiency, and SLC39A13 deficiency (the original spondylo-cheiro-dysplastic type) [3]. Of note, these three conditions are also included in the 2019 revision of the skeletal dysplasia nosology [6]. SLC39A13 deficiency is rare and following the initial eight affected individuals, only a single case report has been published [7]. We present a follow up of two of the original patients [5] as well as clinical, radiological, and genetic findings on four new patients from three families.

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

All affected individuals and/or their legal representatives in the study gave their informed consent to the use of their clinical data, as well as for molecular studies in a diagnostic context (see below).

#### *2.1. Molecular Analysis*

Molecular studies for patients 1 and 2 were done in the Lausanne laboratory as described [5]. Molecular studies in patients 3 to 6 were done for diagnostic purposes with appropriate informed consent from patients and their guardians. The studies were done using routine diagnostic sequencing procedures in certified diagnostic laboratories: Analysis for patients 3, 5, and 6 was done in Lausanne, while the analysis of patient 4 was done in Nijmegen.

#### *2.2. Analysis of Pyridinium Crosslink Products by HPLC*

Spot urine samples were acid hydrolyzed and analyzed by reverse-phase HPLC as previously described [4].

#### *2.3. Analysis of Facial Features*

Anonymized frontal facial photographs of individuals with confirmed biallelic *SLC39A13* variants were used to capture the facial gestalt of SLC39A13 deficiency and to compare it to individuals who are clinically normal. All images were fully de-identified through the use of the DeepGestalt image analysis technology [8]. De-identified data of images from 8 patients (the six patients reported here, plus two of the patients described by Giunta et al. (2008) whose photographs had been submitted to us for clinical consultation) were uploaded to the Face2Gene Research app [9] and matched to controls by age, sex, and ethnicity to produce artificial composite images. Because of the small number of patients, the comparison was run twice [10] with two different sets of matched controls. The comparison and separation quality between the three groups was evaluated by measuring the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. To estimate the statistical power of DeepGestalt in distinguishing affected individuals from controls, a cross-validation scheme was used, including a series of binary comparisons between all groups. For these binary comparisons, the data was split randomly multiple times into training sets and test sets. Each such set contained half of the samples from the group, and this random process was repeated 10 times [8,10]. The results of the binary comparisons were reported both numerically and graphically.

#### *2.4. Clinical Reports*

#### 2.4.1. Individuals 1 and 2 (reported in part by Fukada et al., 2008 [5])

The parents, of Portuguese origin, were of average stature and clinically unremarkable. They were not knowingly related, although molecular workup showed that they shared a common haplotype harboring the *SLC39A13* pathogenic variant [5]. The affected children, a boy and a girl, were born at term from uncomplicated pregnancies and were of normal size and weight at birth but showed progressive short stature beginning in the second half of the first year of life. Among the clinical signs in early childhood were muscular hypotonia and soft skin, leading to the diagnostic suspicion of the Ehlers–Danlos syndrome (EDS). During childhood, the main clinical signs and features were thin, fragile, but not hyperelastic skin that bruised easily and was particularly thin on the hands and feet; varicose veins; moderate joint laxity; blueish or greyish sclerae; down-slanting palpebral fissures; and the absence of one or more teeth in permanent dentition. Both had astigmatism in childhood. Radiographic examination revealed moderate osteopenia, flattened or biconcave vertebral bodies with flaky irregularity of the endplates as well as mild dysplastic changes at the metaphyses of long bones and of the phalanges. At the time of the first report, they were 28 years old and 145 cm tall (individual 1, male); and 20 years old and 135 cm tall (individual 2, female). Their body proportions were normal, indicating that the platyspondyly was accompanied by shortening of the long bones. Their skin remained thin and fragile, and the subcutaneous fat tissue was sparse. Both individuals had marked venous varicosities on their legs and feet. Individual 1 suffered from a cerebral hemorrhage posteriorly to the left putamen at age 22 years, from which he recovered completely. He has successfully completed higher education.

Subsequent to the initial publication in 2008, individual 1 has had no further complications; he is professionally active, has married, and his wife has given birth to a healthy child. Individual 2, his younger sister, had obtained higher education degrees at age 22 and 25 years. At age 25 years, cerebral vascular imaging was obtained, and no abnormalities were observed. At age 26 years, two weeks after the interruption of hormonal contraception prescribed because of irregular cycles, she suffered from arterial thromboembolism that caused cerebral ischemia with right arm paresis and Broca's aphasia. Several days after initiation of aspirin therapy, she developed cerebral hemorrhage leading to complete right hemiparesis. Fortunately, she was able to partially recover over four years on physiotherapy but still has partial function of the right hand and myoclonus on the right arm. She received speech/language therapy; however, she still has some degree of dysphasia. Nevertheless, she was able to return to her previous work.

#### 2.4.2. Individual 3

This is a girl born by caesarean section at 33 weeks of gestation from a dizygotic twin pregnancy. Both the patient and her twin sister were 2500 g at birth. The patient developed progressive short stature over her first years of life, while her twin sister grew normally. Motor and intellectual development were normal. The parents are first cousins of Turkish origin and measure 148 cm and 170 cm with normal proportions. An elder brother had normal growth and body proportions. During her teenage years, she developed severe keratoconus with no complications so far. She complains about back pain and her skin bruises easily, with hematomas and blood blisters at sites of friction. Radiographic examination of the spine at age 15 revealed platyspondyly of the thoracic and lumbar vertebrae with mild anterior beaking. On physical examination at age 23, she had short stature. There was moderate diffuse joint laxity most marked on the finger joints. The skin on her hands is thin and wrinkled and she has tapering fingers with narrow end phalanges. Facial features include downslanting palpebral fissures, a flat face, protuberant eyes with reduced periocular tissue, greyish sclerae, and a small mouth.

#### 2.4.3. Individual 4

This boy was the fourth child from healthy parents of Turkish descent. The parents were second cousins and measured 165 cm (father) and 158 cm (mother). His three older sibs have normal growth. He was born after an uneventful term pregnancy with a birthweight of 3500g. He developed progressive short stature over his first years of life and was first seen by a pediatrician at age 8 1/2 years. Height was 113.4cm (−3.8 SDS) and weight was 23.5kg (−1.5 SDS). Psychomotor development was normal. He complained of knee and ankle pain after exercise. He had mild weakness of the hands, making it difficult to open bottles. He had no clear dysmorphic features, but his palpebral fissures were down-slanting and he had oligodontia (missing five teeth). He has had no eye complication so far. He had a short trunk and stands with lumbar hyperlordosis. He had loose skin on hands and feet, moderate hyperlaxity of fingers and toes, valgus deformity of the ankles, and bilateral pes planus. Radiological examination showed platyspondyly of the thoracic and lumbar spine with mild anterior beaking and mild metaphyseal irregularities especially of the distal ulna. He first received a clinical diagnosis of spondylometaphyseal dysplasia. Molecular analysis of *SHOX*, *FGFR3*, *TRPV4*, and *LTBP3* were negative, but further studies showed a biallelic missense change in *SLC39A13* (Table 1). Additional urine analysis showed a low ratio of pyridinoline-to-deoxypyridinoline (Figure 3).



Notes: \* these two siblings have been reported by us before (Fukada et al., 2008); \*\* as assessed in March 2020 on the gnomAD database (https://gnomad.broadinstitute.org/); \*\*\* as assessed in March 2020 on the VARSOME integrative website (https://varsome.com/).

#### 2.4.4. Individuals 5 and 6

The family originates from Syria and access to early medical records is impossible. The parents, who are first cousins, are clinically healthy and measure 150 cm in height (mother) and 160 cm (father). The proband (patient 6; Figure 1) is a 9-year-old boy referred for evaluation for small stature. We found that his elder sister is similarly affected (patient 5), while a younger sister is clinically normal. Individual 6 was born at term after an uneventful pregnancy, although IUGR was noted and the mother reports reduced fetal movements throughout the pregnancy. Birth data were not recorded, but the mother remembers his length to be reduced. At birth, the mother noted the he had no spontaneous movements, had edema on both feet, and increased palmar creases on his hands and feet. Throughout childhood, psychomotor development was slightly delayed, but intelligence and behavior at age 9 years were normal. At this age, his growth parameters were height 111 cm (−3.9 SD), weight 19 kg (−3.3 SD), and head circumference 51.5 cm (−0.78 SD). He had down-slanting palpebral fissures, protruding eyes, hypertelorism, and a large forehead. An ophthalmologic evaluation revealed bilateral myopia and marked keratoconus. Hearing was normal. Four teeth of the permanent dentition were found to be missing. His elbow joints appeared prominent, but he had no other signs of joint deformity or scoliosis. He had hyperlaxity of small joints and complained occasionally of joint pain particularly in his knees. The skin was thin and wrinkled, particularly on the palmar side of his hands and feet, and the thenar muscles were underdeveloped. Radiographic studies showed small epiphyses and posterior subluxation of the radii and mild platyspondyly. Routine blood tests were unremarkable. Analysis of *SHOX* done prior to referral was negative.

**Figure 1.** Clinical and radiographic findings. Individuals 5 (panels **A** and **B**) and 6 (panels **A** and **C**) at the age of eight and ten years. Panels **D** and **E**: missing upper incisives, and missing lower incisives with persistence of primary teeth in individuals 6 at age 9 years. Panel **F**: foot of individual 6 showing finely wrinkled skin and abnormally deep furrows. Panel **G**: Palmar aspects of the hands of individual 5 showing thin skin with increased number of fine wrinkles and mild thenar and hypothenar atrophy. Panel **H** shows the hand radiograph of individual 5 at age ten years. There is mild diaphyseal overconstriction of radius and ulna, metacarpals, and phalanges; bone maturation is roughly appropriate; overall, the changes are mild and non-diagnostic. Panels **I** and **K**: lateral spine radiographs of individual 3. At age 6 years, there is moderate platyspondyly. At age 15, the end plates have a marked concave conformation.

Individual 5 is the older sister of individual 6. According to the birth date declared at immigration, she would be 10 years old. However, we doubt the accuracy of this date and suspect that she is older than the stated age. Birth parameters have not been recorded. Her growth parameters were as follows: height 131.5 cm (−1 SD for age 10), weight 29 kg (−0.81 SD for age 10), and head circumference 51.5 cm (−0.55 SD for age 10). Shortly after she was referred, she had menarche and her growth has slowed down significantly. She had protruding eyes and a large forehead, but her facial appearance was less conspicuous than that of her brother or other affected individuals. She had myopia, and four

incisor teeth were missing. Intelligence and hearing were normal. She had long tapering fingers and her interphalangeal joints were hyperextensible. The skin was thin and wrinkled, particularly on the palmar side of her hands and feet.

#### **3. Results**

Figure 1 shows some clinically relevant clinical and radiographic features in SLC39A13-deficient individuals.

Table 1 shows the genetic variants identified in *SLC39A13*. Of note, the variants were homozygous in all patients, underlining the rarity of the disorder.

Figure 2 shows the results of the automated DeepGestalt technology. Both comparisons (*SLC39A13* group vs. control group 1, and *SLC39A13* group vs. control group 2) gave highly significant results (AUC= 0.991 p = 0.006 and AUC= 0.986 *p* = 0.004).

**Figure 2.** DeepGestalt technology analysis of facial features in SLC39A13 deficiency. The upper part shows the "averaged" artificial composite facial gestalt image of individuals with SLC39A13 deficiency (panel **A**) vs. two matched control groups (panels **B** and **C**). The lower part shows the result of the comparisons between **A** vs. **B**, and **A** vs. **C**.

Figure 3 shows the ratio of the collagen crosslink derivates, pyridinoline and deoxypyridinoline, in the urine of five patients and five roughly age-matched controls. The ratio of pyridinoline-todeoxypyridinoline is reduced in the SLC39A13-deficient subjects, indicating a relative deficiency of the hydroxylated form, pyridinoline, relative to the non-hydroxylated form, deoxypyridinoline. The values clearly segregated in two distinct clusters with no overlap. In addition to confirming the notion of reduced lysyl hydroxylation as a pathogenic mechanism in these patients, the findings show that determination of urinary crosslink derivates can be a useful diagnostic screening method. Anecdotally, because of the role of vitamin C in collagen hydroxylation, individuals 1 and 2 took an oral dose of 1000 mg of vitamin C over four weeks and sent us urine samples for analysis before and twice during the test period. Unfortunately, there was no change in the pyridinoline-to-deoxypyridinoline ratio.

**Figure 3.** Ratio of the collagen crosslink derivates, pyridinoline and deoxypyridinoline, in the urine of five SLC39A3-dificient individuals and in five roughly age-matched controls. The ratio is consistently lower in SLC39A13-deficient individuals, with no overlap to the control group.

#### **4. Discussion**

#### *4.1. Clinical Aspects of SCL39A13 Deficiency*

Our observations confirm the previous findings of Giunta et al. (2008), Fukada et al. (2008), and Dusanic et al. (2018) that SLC39A13 deficiency is associated with a clinical phenotype characterized by post-natal short stature, connective tissue weakness affecting mainly the skin and peripheral joints, a characteristic facial appearance, and a moderate skeletal dysplasia. These additional observations help to delineate a more precise phenotype.

**Muscular hypotonia**—This has been seen in several *SLC39A13* patients in the neonatal period and in early childhood; in the patient described by Dusanic et al. (2018), muscular hypotonia was conspicuous enough to lead to investigations for myopathy with some abnormal (though non-specific) findings in a muscle biopsy at adolescence. While the prevalence of myopathy in SLC39A3 deficiency and its structural features remain to be investigated, the observations confirm the well-known phenomenon that connective tissue diseases, and particularly EDS type VI-A (lysyl hydroxylase deficiency), may present as the "floppy infant" phenomenon and/or with hypotonia with delayed motor development.

**Growth and stature**—postnatal reduced linear growth seems to be a salient and consistent feature of SLC39A13 deficiency; this is in contrast to other types of connective tissue disease and EDS types, where short stature is not a prominent feature. The two other forms of "spondylodysplastic EDS", namely B3GALT6 deficiency and B4GALT7 deficiency, are also associated with short stature, but in

those conditions, short stature can be explained by skeletal dysplasia. In *SLC39A13*, body proportions are normal and short stature seems to be more intrinsic and represents a true growth failure and not a consequence of bone dysplasia.

**Eye findings**—Myopia has been reported in several *SLC39A13* patients and also seems to be a direct (though non-specific) manifestation of the primary genetic defect. This may be connected to the thin sclerae that often have a blueish or greyish color, particularly in individuals with darker pigmentation. Keratoconus, as seen in three patients, is potentially dangerous because of the possibility of perforation or rupture; ophthalmologic investigation should be recommended in all confirmed cases.

**Dental features**—Absence of one or several incisor teeth in the permanent dentition has been observed in a majority of SLC39A13-deficient individuals. In our case 6, the lower incisors of the primary dentition were also small and dysplastic. Hypodontia and oligodontia are consistent, if perhaps not obligate, features of the disorder.

**Vascular complications**—marked varicosities of the lower legs have been described in a number of *SLC39A13* patients. Clinically, more importantly, the two elder patients known (patients 1 and 2 in Fukada et al. (2008) and in this study) have both suffered cerebral hemorrhage. While this may not be statistically significant and may even be causally unrelated (e.g., hormonal contraception in patient 2), it is worth keeping in mind as a potential complication. Fortunately, other major vascular events have thus far not been recorded.

**Facial features**—analysis of the facial features with the DeepGestalt technology and comparison with age-matched controls confirms that there is an SLC39A13-associated facial phenotype that is significantly divergent from that of control individuals. This phenotype includes a rather flat face, mild hypertelorism, downslanting palpebral fissures, lack of periocular connective tissue giving the impression of prominent eyes, and a small mouth. The Face2Gene system evoked a few other syndromes with some resemblance to the *SLC39A13* phenotype, such as Noonan syndrome (downslanting palpebral fissures), Stickler syndrome (flat face; associated with *COL2A1* haploinsufficiency), as well as—interestingly—the vascular type of EDS (EDS IV; dominant *COL3A1*), probably because of the "hollow eyes". Thus, analogy between SLC39A13 deficiency and vascular EDS may include vascular fragility as well as a moderate resemblance of facial features.

#### *4.2. What Are the Pathogenetic Mechanisms Leading from SLC39A13 Loss of Function to the Complex Clinical Phenotype?*

The pathogenesis of SLC39A13 deficiency remains poorly understood. The observation of reduced collagen hydroxylation indicates that a partial failure of collagen crosslinking is one pathogenic mechanism [4], and the determination of pyridinoline-to-deoxypyridinoline is a useful diagnostic help. As correctly indicated by Giunta et al. (2008), the hydroxylation defect probably involves collagens other than collagen 1, such as collagen 2 and collagen 3, and this may explain the mild chondrodysplastic features (collagen 2) as well as the thin and fragile skin and the putative vascular fragility (collagen 3). However, it is difficult to ascribe the short stature, facial features, and oligodontia solely to reduced collagen hydroxylation. The hypothesis that part of the pathogenesis may involve the role of zinc as an intracellular messenger [5] may be justified, but experimental evidence is lacking; in particular, SLC39A13 individuals do not show signs of immune system dysfunction in contrast with the mouse model [5,11].

#### *4.3. Diagnosis*

From the diagnostic perspective, presentation with short stature and dysmorphic facial features may lead pediatricians and clinical geneticists to classify these patients within the group of genetic syndromes rather than connective tissue disorders; this "syndromic" presentation of SLC39A13 is not well known and this report may help raise awareness of this condition and its presenting features among clinical geneticists and dysmorphologists.

**Author Contributions:** Conceptualization, C.K., S.U., and A.S.-F.; methodology, A.S.-F., M.K., B.C.-X. and N.F.; software, N.F. and A.S.-F.; formal analysis, B.C.-X., M.K., N.F. and A.S.-F.; investigation, C.K., Y.H.-H., M.C. and A.S.-F.; writing—original draft preparation, C.K. and A.S.-F.; writing—review and editing, C.K., B.C.-X., C.M., M.K., S.U. and A.S.-F.; visualization, A.S.-F.; supervision, A.S.-F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We wish to express our gratitude to the individuals who participated in this study. Through the generous sharing of their data, they will help in improving the diagnosis and hopefully the treatment of other affected children and adults. Individual 3 originally came to our observation (YHH, SU, AS-F) through the former European Skeletal Dyplasia Network (ESDN). This study was also supported by a generous donation by the Fondation Guillaume Gentil (Lausanne) to the Division of Genetic Medicine of the Lausanne University Hospital (CHUV).

**Conflicts of Interest:** Superti-Furga is a member of scientific advisory board of FDNA (non-paid).

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

#### *Review*

### **Recent Advances in the Pathophysiology of Musculocontractural Ehlers-Danlos Syndrome**

### **Tomoki Kosho 1,2,3,\*, Shuji Mizumoto 4, Takafumi Watanabe 5, Takahiro Yoshizawa 6, Noriko Miyake <sup>7</sup> and Shuhei Yamada <sup>4</sup>**


Received: 12 December 2019; Accepted: 23 December 2019; Published: 29 December 2019

**Abstract:** Musculocontractural Ehlers–Danlos Syndome (mcEDS) is a type of EDS caused by biallelic pathogenic variants in the gene for carbohydrate sulfotransferase 14/dermatan 4-*O*-sulfotransferase 1 (*CHST14*/*D4ST1*, mcEDS-*CHST14*), or in the gene for dermatan sulfate epimerase (*DSE*, mcEDS-*DSE*). Thus far, 41 patients from 28 families with mcEDS-*CHST14* and five patients from four families with mcEDS-*DSE* have been described in the literature. Clinical features comprise multisystem congenital malformations and progressive connective tissue fragility-related manifestations. This review outlines recent advances in understanding the pathophysiology of mcEDS. Pathogenic variants in *CHST14* or *DSE* lead to reduced activities of relevant enzymes, resulting in a negligible amount of dermatan sulfate (DS) and an excessive amount of chondroitin sulfate. Connective tissue fragility is presumably attributable to a compositional change in the glycosaminoglycan chains of decorin, a major DS-proteoglycan in the skin that contributes to collagen fibril assembly. Collagen fibrils in affected skin are dispersed in the papillary to reticular dermis, whereas those in normal skin are regularly and tightly assembled. Glycosaminoglycan chains are linear in affected skin, stretching from the outer surface of collagen fibrils to adjacent fibrils; glycosaminoglycan chains are curved in normal skin, maintaining close contact with attached collagen fibrils. Homozygous (*Chst14*−/−) mice have been shown perinatal lethality, shorter fetal length and vessel-related placental abnormalities. Milder phenotypes in mcEDS-*DSE* might be related to a smaller fraction of decorin DS, potentially through residual DSE activity or compensation by DSE2 activity. These findings suggest critical roles of DS and DS-proteoglycans in the multisystem development and maintenance of connective tissues, and provide fundamental evidence to support future etiology-based therapies.

**Keywords:** musculocontractural Ehlers–Danlos Syndome; carbohydrate sulfotransferase-14 (CHST14)/dermatan 4-*O*-sulfotransferase-1 (D4ST1); *CHST14*; dermatan sulfate epimerase (DSE); *DSE*; dermatan sulfate (DS); decorin; collagen

#### **1. Introduction**

Musculocontractural Ehlers–Danlos Syndome (mcEDS) is a type of EDS, caused by biallelic pathogenic variants in the gene for carbohydrate sulfotransferase 14/dermatan 4-*O*-sulfotransferase 1 (*CHST14*/*D4ST1*, mcEDS-*CHST14*) (MIM#601776), or in the gene for dermatan sulfate epimerase (*DSE*, mcEDS-*DSE*) (MIM#615539) [1–3]. mcEDS-*CHST14* was originally described as three independent conditions: A rare type of arthrogryposis syndrome "adducted thumb-clubfoot syndrome" [4]; a specific type of EDS "EDS, Kosho type" [5,6]; and a subset of kyphoscoliosis type without lysyl hydroxylase deficiency [7]. To date, 41 patients from 28 families have been reported to have mcEDS-*CHST14* [4–22]. mcEDS-*DSE* was identified in a patient with a phenotype similar to that of patients with mcEDS-*CHST14* [23], as well as in four additional patients from three families [18,24]. These disorders were defined as subtypes of EDS, based on the International Classification of the EDSs [3]. Clinical features are highly characteristic, comprising multisystem congenital malformations such as craniofacial features (e.g., large fontanelle, hypertelorism, short and downslanting palpebral fissures, blue sclerae, short nose with hypoplastic columella, low-set and rotated ears, high palate, long philtrum, thin upper lip vermilion, small mouth and micro-retrognathia), multiple congenital contractures (e.g., adduction–flexion contractures of thumbs and talipes equinovarus), and visceral and ocular malformations. Features also include progressive connective tissue fragility-related manifestations, such as skin hyperextensibility, bruisability, and fragility with atrophic scars; recurrent dislocations; progressive talipes or spinal deformities; pneumothorax or pneumohemothorax; large subcutaneous hematomas; and/or diverticular perforation (Figure 1) [1,2]. Major diagnostic criteria of the disorder are as follows: 1) congenital multiple contractures, characteristically adduction–flexion contractures and/or talipes equinovarus (clubfoot); 2) characteristic craniofacial features, which are evident at birth or in early infancy; 3) characteristic cutaneous features including hyperextensibility, bruisability and fragility with atrophic scars, as well as increased palmer wrinkles [3]. Minor criteria as follows: 1) recurrent/chronic dislocations, 2) pectus deformities (e.g., flat or excavated), 3) spinal deformities (e.g., scoliosis or kyphoscoliosis), 4) peculiar fingers (e.g., tapered, slender, or cylindrical), 5) progressive talipes deformities (e.g., valgus, planus, or cavum), 6) large subcutaneous hematomas, 7) chronic constipation, 8) colonic diverticula, 9) pneumothorax/pneumohemothorax, 10) nephrolithiasis/cystolithiasis, 11) hydronephrosis, 12) cryptorchidism in males, 13) strabismus, 14) refractive errors (e.g., myopia or astigmatism) and/or 15) glaucoma/elevated intraocular pressure [3].

In this review, we describe the comprehensive pathophysiological findings of mcEDS, as demonstrated in previous studies including our recent reports.

**Figure 1.** *Cont.*

**Figure 1.** Clinical photographs and radiological images of patients with mcEDS-*CHST14*. Clinical photographs of a patient with heterozygous variants Pro281Leu/Try293Cys at age 23 days (**A**), 3 years (**B**), and 16 years (**C**,**D**); those of a patient with a homozygous variant Pro281Leu at age 2 months (**H**,**M**,**N**), 3 months (**I**), 6 years (**O**,**P**), and 28 years (**J**,**Q**); photographs of a patient with a homozygous variant "P281L" in the neonatal period (**E**) and at age 30 years (**F**,**G**,**T**); and photographs of a patient with heterozygous variants Pro281Leu /Cys289Ser at age 1 month (**K**), 16 years (**V**), and 19 years (**L**,**R**,**S**) [5]. Radiological image of a patient with heterozygous variants Pro281Leu /Try293Cys at age 6 years (**U**); images of a patient with a homozygous variant Pro281Leu at age 28 years (**W**–**Z**,**a**) [5,13]. (U, reproduced from Kosho et al. *Am. J. Med. Genet. Part A* **2005**, *138A*, 282–287, with permission from Wiley-Liss, Inc.; the other images, reproduced from Kosho et al. *Am. J. Med. Genet. Part A* **2010**, *152A*, 1333–1346, with permission from Wiley-Liss, Inc.).

#### **2. Molecular Findings**

Pathogenic variants have been detected throughout *CHST14* (NM\_130468.4): 11 missense variants, five frameshift variants, and three nonsense variants in patients with mcEDS [4,6,7,14–19,22] (Figure 2). The p.(Pro281Leu) variant is most common (*n* = 10 families), followed by p.(Try293Cys) (*n* = 4), p.(Val49\*) (*n* = 3), p.(Arg213Pro), and p.(Phe209Ser) (*n* = 2); p.(Arg29Glyfs\*113), p.(Lys69\*), p.(Gln113Argfs\*14), p.(Arg135Gly), p.(Leu137Gln), p.(Cys152Leufs\*10), p.(Arg218Ser), p.(Gly228Leufs\*13), p.(Glu262Lys), p.(Tyr266\*), p.(Arg274Pro), p.(Met280Leu), p.(Cys289Ser), p.(Trp327Cysfs\*29), and p.(Glu334Glyfs\*107) variants are particularly uncommon (*n* = 1 for all). Furthermore, p.(Gly19Trpfs\*19) and p.(Lys26Alafs\*16) variants have been detected in patients with features similar to those of mcEDS, among patients with hereditary connective tissue disorders and skeletal dysplasia, respectively (Figure 2A) [25,26]. Three missense variants have been detected in *DSE* (NM\_013352.4): p.(Arg267Gly), p.(Ser268Leu), and p.(His588Arg); and one frameshift variant, p.(Pro384Trpfs\*9), has also been detected (Figure 2B) [18,23,24]. Another frameshift variant, p.(Gly216Glufs\*3), was detected in a patient with features similar to those of mcEDS, among patients with chondrodysplasia who exhibited multiple dislocations (Figure 2B) [27].

**Figure 2.** Published pathogenic protein changes of D4ST1 and DSE in mcEDS. (**A**) Previously published truncating and non-truncating protein alterations in D4ST1 are shown in upper and lower panels, respectively. Black box indicates 5 -phosphosulfate binding site and gray box indicates 3 -phosphate binding site. (**B**) Previously published truncating and non-truncating protein alterations in DSE are shown in upper and lower panels, respectively. Black box indicates the signal peptide.

No apparent genotype-phenotype correlations have been reported in patients with mcEDS-*CHST14*. Phenotypes of patients with mcEDS-*DSE* seem to be milder than those of patients with mcEDS-*CHST14* [18,24].

#### **3. Glycobiological Findings**

Normal, biosynthetic pathways of chondroitin sulfate (CS) and dermatan sulfate (DS) are shown in Figure 3A. Reduced activity of D4ST1 in fibroblast cultures of skin from a patient with mcEDS-*CHST14* caused by compound heterozygous p.(Pro281Leu)/(Tyr293Cys) substitutions in *CHST14*, as well as in fibroblasts from a patient with mcEDS-*CHST14* caused by a homozygous p.(Pro281Leu) substitution in *CHST14*, showed a marked reduction in D4ST1 activity (Figure 3B); this change in activity resulted in a negligible amount of DS and an excessive amount of CS (Figure 3C) [6].

Decorin, which consists of a core protein and a single glycosaminoglycan (GAG) chain, is a major DS-proteoglycan (PG) that plays an important role in the assembly of collagen fibrils in the skin [28]; it also plays roles in the pathophysiology of mcEDS-*CHST14* [1–4,6,7]. GAG chains of decorin-PG from skin fibroblasts of a patient with p.(Pro281Leu)/(Tyr293Cys) substitutions, as well as from skin fibroblasts of a patient with a homozygous p.(Pro281Leu) substitution, contained only CS and no DS; in contrast, GAG chains of decorin-PG from skin fibroblasts of healthy controls contained mainly DS (Figure 3D) [6,18]. 4-*O*-Sulfation in CS and DS chains functions as an inhibitor of DSE [29]. Thus, impaired 4-*O*-sulfation inhibition due to D4ST1 deficiency enables back-epimerization from L-iduronic acid (IdoUA) to D-glucuronic acid (GlcUA) (Figure 3E) [4,6,7]. In our laboratory, we have established a urinary disaccharide analysis of CS/DS chains through an anion-exchange chromatography after treatment with DS-specific degrading enzymes; this analysis method showed that no DS was present in the urine of eight patients with mcEDS-*CHST14* [30]. This result suggested a systemic depletion of DS in patients with mcEDS-*CHST14*; thus we presume that our urinary disaccharide analysis method can be implemented to allow a non-invasive screening for mcEDS-*CHST14* [30].

**Figure 3.** Biosynthesis of CS/DS and glycobiological analysis in mcEDS-*CHST14*. (**A**) Biosynthetic assembly of CS and DS chains by glycosyltransferases, epimerases, and sulfotransferases. It starts with the biosynthesis of a tetrasaccharide linker region, glucuronic acid-β1,3-galactose-β1,3-galactose-β1,4-xylose-β1-*O*-(GlcUA-Gal-Gal-Xyl-), onto serine residues of specific core proteins of PGs by β-xylosyltransferase (XylT), GalT-I, GalT-II and β1,3-glucuronosyltransferase-I (GlcAT-I), respectively. Subsequently, a repetitive disaccharide region [*N*-acetyl-d-galactosamine(GalNAc)-GlcUA)]n of chondroitin is elongated by the actions of *N*-acetyld-galactosaminyltransferase-I (GalNAcT-I), *N*-acetyl-d-galactosaminyltransferase-II (GalNAcT-II) and glucuronyltransferase-II (GlcAT-II), which are encoded by CS *N*-acetylgalactosaminyltransferase-1 and -2, chondroitin synthase-1, -2, and -3 and chondroitin polymerizing factor genes. Chondroitin chains are matured to CS through modifications by chondroitin 4-*O*-sulfotransferase (C4ST), chondroitin 6-*O*-sulfotransferase (C6ST) and uronyl 2-*O*-sulfotransferase (UST). A disaccharide-repeating region of dermatan is synthesized through epimerization of a carboxyl group at C5 from GlcUA to IdoUA by DSE. A mature DS chain is synthesized through modification with sulfation by D4ST1 and UST. (**B**) Sulfotransferase activity toward dermatan in fibroblast cultures of skin from two patients with mcEDS-*CHST14* (patient 12 with heterozygous variants Pro281Leu/Tyr293Cys; patient 14 with a homozygous variant Pro281Leu), the mother of patient 12, and a sex- and age-matched healthy volunteer [6]. \* *p* < 0.0001 by two-tailed unpaired *t*-test. (**C**) Total amounts of CS and DS derived from fibroblast cultures of skin [6]. Total disaccharide contents of CS (white box) and DS (black box) were calculated based on the peak area in the chromatograms of the digests with chondroitinase AC and chondroitinase B, respectively. (**D**) Proportion of the disaccharide units in the CS-DS hybrid chain in decorin-PGs secreted by the fibroblasts [6]. Cyan, green, red and orange boxes are GlcUA-GalNAc(4S), GlcUA-GalNAc(6S), IdoUA-GalNAc(4S), and IdoUA(2S)-GalNAc(4S), respectively. Abbreviations of 2S, 4S and 6S indicate 2-*O*-, 4-*O*- and 6-*O*-sulfate, respectively. (**E**) Schematic diagram of the biochemical mechanism in the replacement of DS by CS in mcEDS-*CHST14*. Defect in D4ST1 enables a back-epimerization reaction that converts IdoUA back to GlcUA to form chondroitin by DSE, followed by the 4-*O*-sulfation and/or 6-*O*-sulfation of GalNAc residues in chondroitin by C4ST1 and C6ST1, respectively. (B–D, reproduced from Miyake et al. *Hum. Mutat.* **2010**, *31*, 1233–1239, with permission from Wiley-Liss, Inc.).

Regarding patients with mcEDS-*DSE*, reduced activity of DSE in fibroblast cultures of skin from a patient with a homozygous p.(Ser268Leu) substitution resulted in marked reduction of DS disaccharides, compared with healthy controls [23]. The total amount of CS in the cell fraction from affected skin fibroblasts was increased by approximately 1.5-fold, which might reflect increased synthesis and/or reduced conversion of CS chains [23]. A minor fraction of DS from decorin-PG was present in skin fibroblasts from a patient with a homozygous p.(Ser268Leu) substitution [23]; this suggested residual DSE activity or compensation by DSE2, which might be related to milder phenotypes in patients with mcEDS-*DSE* than in patients with mcEDS-*CHST14* [18].

#### **4. Pathological Findings**

The pathology of mcEDS-*CHST14*, as the simplest model for complete depletion of DS, has been extensively investigated using affected skin specimens. Light microscopy of skin specimens (hematoxylin and eosin staining) from patients with compound heterozygous p.(Pro281Leu)/(Cys289Ser) or p.(Pro281Leu)/(Tyr293Cys) substitutions showed that fine collagen fibers were predominantly present in the reticular to papillary dermis; marked reduction of normally thick collagen bundles were also observed (Figure 4A) [6]. Immunohistochemistry staining of decorin core protein in skin specimens from patients with compound heterozygous p.(Pro281Leu)/(CYs289Ser) or p.(Pro281Leu)/(Tyr293Cys) substitutions showed that decorin core protein was present on collagen fibers that were thin and filamentous without clear boundaries; in contrast, skin specimens from healthy controls showed decorin core protein on collagen fibers that were thick bundles with clear boundaries [31] (Figure 4B). Transmission electron microscopy of skin specimens from five patients with compound heterozygous p.(Pro281Leu)/(Cys289Ser), p.(Pro281Leu)/(Tyr293Cys), or p.(Phe209Ser)/(Pro281Leu) substitutions showed that collagen fibrils were dispersed in the papillary to reticular dermis, whereas skin specimens from healthy controls exhibited collagen fibrils that were regularly and tightly assembled (Figure 4C) [6,31]. Transmission electron microscopy-based cupromeronic blue staining to visualize GAG chains on affected skin samples showed that GAG chains were linear, stretching from the outer surface of collagen fibrils to adjacent fibrils, whereas skin samples from healthy controls exhibited curved GAG chains that maintained close contact with attached collagen fibrils (Figure 4D) [31]. This structural alteration of GAG chains of decorin is presumably related to the biochemical alteration from DS to CS: the structure of DS-GAG chains is flexible because L-IdoUA residues in DS can easily adopt any of the nearly equi-energetic 1C4, 2S0, and 4C1 conformations, whereas the structure of CS-GAG chains is rigid because D-GlcUA in CS only adopts the 4C1 conformation [6,32,33].

Furthermore, focused ion beam scanning electron microscopy using cupromeronic blue staining uncovered the structure of collagen fibrils in association with GAG chains (likely comprising decorin): GAG chains form a ring mesh-like structure with each ring surrounding a collagen fibril at its D band, fusing with adjacent rings to form a planar network [34]. Abnormally stretching CS-GAG chains of decorin in the affected skin would disrupt the ring-mesh structure of collagen fibrils (Figure 4E), which could result in substantial fragility.

**Figure 4.** Skin pathology of mcEDS-*CHST14*. (**A**) Light microscopy (hematoxylin and eosin staining). In the skin specimen from a patient with heterozygous variants Pro281Leu/ Cys289Ser (panel p), fine collagen fibers are present predominantly in the reticular to papillary dermis with marked reduction of thick collagen bundles; thick collagen bundles are observed in a skin specimen from a healthy control volunteer (panel c) [6]. (**B**) Immunohistochemical staining for decorin core protein. Decorin core protein is present on collagen fibers in thick bundles in a skin specimen from a healthy control volunteer (panel c), but on thin and filamentous collagen fibers without clear boundaries in a skin specimen from a patient with heterozygous variants Pro281Leu/ Cys289Ser (panel p) [31] (**C**) Transmission electron microscopy. Collagen fibrils are regularly and tightly assembled in a skin specimen from a healthy control volunteer (panel c), but are dispersed in the papillary to reticular dermis in skin specimens from a patient with heterozygous variants Pro281Leu/ Tyr293Cys (panel p1)

and a patient with a novel homozygous variant (panel p2) [31]. (**D**) Transmission electron microscopy-based cupromeronic blue staining. GAG chains are curved and maintain close contact with attached collagen fibrils in the skin specimens from a healthy control volunteer (panels c-a, c-b); conversely, they are linear and stretch from the outer surface of collagen fibrils to adjacent fibrils in skin specimens from a patient with heterozygous variants Pro281Leu/ Tyr293Cys (panel p1), a patient with a novel homozygous variant (panels p2a and p2b) and another patient with heterozygous variants Pro281Leu/ Tyr293Cys (panel p3) [31]. (**E**) Schematic representations of collagen fibrils and GAG chains. Decorin core protein binds to D bands of collagen fibrils both in normal skin and in affected skin. GAG chains composed of DS adhere to collagen fibrils along D bands, beginning from the core protein (panels a and c), whereas GAG chains composed of CS extend linearly and perpendicularly to collagen fibrils from the core protein (panels b and d) [31]. Putative spatial disorganization of collagen fibril networks in the skin of patients (panel e) (A, reproduced from Miyake et al. *Hum. Mutat.* **2010**, *31*, 1233–1239, with permission from Wiley-Liss, Inc.; B–E, reproduced from Hirose et al. *Biochim. Biophys. Acta Gen. Subj.* **2019**, *1863*, 623–631, with permission from Elsevier, Inc.).

#### **5. Animal Model-Based Findings**

Knockout (*Chst14*−/−) mice were generated through homologous recombination that targeted the only coding exon 1 (i.e., exon 1) of *Chst14* [35,36]. F2 mice showed reduced weight and/or length and reduced bone volume/thickness/density of their lumbar vertebrae [35]. *Chst14*−/<sup>−</sup> mice showed reduced neurogenesis and diminished proliferation of neural stem cells, accompanied by increased expression of glutamate aspartate transporter and epidermal growth factor, compared with findings in both wild-type (*Chst14*+/+) mice and other knockout (*Chst11*−/−) mice, a model for chondroitin 4-*O*-sulfotransferase 1 (C4ST1) deficiency [36]. *Chst14*−/<sup>−</sup> mice also had a smaller body mass (Figure 5A, B), reduced fertility, a kinked tail, and increased skin fragility compared with their wild-type (*Chst14*+/+) littermates; however, brain weight and gross anatomy were not affected [36,37]. Schwann cells from *Chst14*−/<sup>−</sup> mice formed longer processes in vitro and exhibited greater proliferation than those from *Chst14*+/+ mice. Functional recovery and axonal regrowth in *Chst14*−/<sup>−</sup> mice were initially accelerated after femoral nerve transection and suture; after 3 months, these characteristics were similar to those in *Chst14*+/+ littermates. These findings suggested that DS, synthesized by Chst14/D4st1, might be of limited importance for neural development; moreover, it might contribute to the regeneration-restricting environment in the adult mammalian nervous system [38]. Only a few adult *Chst14*−/<sup>−</sup> mice were generated because of perinatal lethality in most of the homozygous mice; these adult *Chst14*−/<sup>−</sup> mice showed significantly shorter crown-rump length, compared with wild-type or heterozygous mice (Figure 5A–C) [37,38]. The placentas of *Chst14*−/<sup>−</sup> fetuses showed a reduced weight, alterations in the vascular structure and ischemic and/or necrotic-like changes (Figure 5D–G). Transmission electron microscopy of homozygous placentas demonstrated an abnormal capillary basement membrane structure in the placental villus, compared with wild-type or heterozygous placentas (Figure 5H–J). These findings showed that DS was essential for placental vascular development and perinatal fetal survival. In addition, DS was suggested to be related to the structure and/or function of capillary basement membranes, which constitutes an extracellular matrix [37].

C4st1, encoded by *Chst11*, transfers a sulfate group from 3 -phosphoadenosine 5 -phosphosulfates to the C4 position of GalNAc residues in chondroitin (Figure 3A). *Chst11*-mutant mice died within 6 hours of birth due to respiratory failure, severe dwarfism and chondrodysplasia (i.e., abnormalities in the cartilage growth plate and chondrocyte columns) [39]. Furthermore, marked reductions in the content and 4-*O*-sulfation of CS, as well as the downregulation of bone morphogenetic protein signaling and upregulation of transforming growth factor-β, have also been observed in *Chst11*-mutant mice. These findings suggest that Chst11/C4st1 and the 4-*O*-sulfation of CS chains are essential for early development and for bone morphogenetic protein and transforming growth factor-β in signaling pathways in cartilage. When considered in the context of *Chst14*−/<sup>−</sup> mouse phenotypes, the above findings suggest that 4-*O*-sulfation is required for the maturation of both CS and DS; moreover, CS and DS exert distinct functions in the development of cartilage and skin, respectively. However, the DS content in *Chst11*-mutant mice remains to be elucidated [39].

**Figure 5.** Mouse model for mcEDS-*CHST14*. (**A**) Numbers (percentages) of each type at embryonic (E18.5) and postnatal (adult) ages. Markedly greater numbers of homozygous fetuses are observed, compared with homozygous adults. (**B**) General appearances of wild-type (*Chst14*+/+), heterozygous (*Chst14*+/−) and homozygous (*Chst14*−/−) fetuses [37]. (**C**) Crown-rump length of each type of fetus (bar: 5 mm). Homozygous mice demonstrate significantly shorter crown-rump length, compared with wild-type or heterozygous mice [37]. (**D**) Appearance of each type of placenta (bar: 5 mm) [37]. (**E**) Homozygous placentas exhibit appearances indicative of hypoxia (left) and necrosis (right) [37]. (**F**) Microphotographs of the chorionic plate side of the placentas (bar: 1 mm). Homozygous placentas demonstrate smaller vascular diameters, compared with wild-type or heterozygous placentas [37]. (**G**) Weight of each type of placenta (mean + standard error of the mean). (**H**) Transmission electron microscopy of capillary basement membrane in the labyrinth zone of each type of placenta (bar: 200 nm). Arrows indicate capillary basement membrane [37]. (**I**) Homozygous placentas show a significantly

thinner capillary basement membrane, compared with wild-type or heterozygous placentas [37]. (**J**) Homozygous placentas show structural abnormalities (arrows) in capillary basement membrane (bar: 0.2 μm) [37]. \* *p* < 0.05, \*\* *p* < 0.01, compared with wild-type; # *p* < 0.05, ## *p* < 0.01, compared with heterozygous; one-way analysis of variance (ANOVA) followed by the Tukey–Kramer post hoc test. (A–J, reproduced from Yoshizawa et al. *Glycobiology* **2018**, *28*, 80–89, with permission from Oxford University Press, Inc.).

Biosynthesis of DS requires D4st1, dermatan sulfate epimerase 1 (DS-epi1) and dermatan sulfate epimerase 2 (DS-epi2), which are encoded by *Chst14*, *Dse* and *Dsel*, respectively [40–43]. Furthermore, D4ST1 interacts directly with DS-epi1, but not with DS-epi2, to form a hetero-complex that is required for the formation of IdoUA blocks in DS chains [44]. *DS-epi1*−/<sup>−</sup> mice showed greater skin fragility compared with wild-type littermates, due to altered collagen fibril morphology [45]; this phenotype was similar to that of *Chst14*−/<sup>−</sup> mice [38]. The numbers of IdoUA blocks were dramatically reduced in DS side chains of decorin-PG and biglycan-PG from the skin of *DS-epi1*−/<sup>−</sup> mice [45], which suggests that DS-epi1 mainly synthesizes IdoUA blocks in vivo. *DS-epi1*−/<sup>−</sup> embryos and newborn mice showed kinked tails, which is a common feature in *Chst14*−/<sup>−</sup> mice [38]; the *DS-epi1*−/<sup>−</sup> embryos and newborn mice also showed significantly thicker epidermal layers through histological staining, compared with heterozygous or wild-type littermates [45,46]. Immunohistochemical staining of epidermal layers in *DS-epi1*−/<sup>−</sup> newborns showed increased expression of keratin 5 in the basal layer and keratin 1 in the spinous layer [46]. Furthermore, a small portion of *DS-epi1*−/<sup>−</sup> embryos showed an abdominal wall defect with herniated intestines, exencephaly and spina bifida. Defective collagen structure in the dermis and imbalanced keratocyte maturation were presumed to cause developmental defects in *DS-epi1*−/<sup>−</sup> mice [46]. These observations indicate that *DS-epi1*−/<sup>−</sup> mice may constitute a useful model of mcEDS-*DSE*. *DS-epi2*−/<sup>−</sup> mice displayed no significant defects and DS-epi1 compensated for the absence of DS-epi2 in most tissue, which indicated that DS-epi1 is the major contributor of epimerase activity [47]. DS-epi2 exhibits higher expression than DS-epi1 in developing the mouse brain [48]. Although CS/DS chains in the brains of *DS-epi2*−/<sup>−</sup> newborn mice demonstrated a 38% reduction in IdoUA content, compared with wild-type littermates, the brains of adult knockout mice showed normal extracellular matrix features [47]. *DS-epi1*−/−/*DS-epi2*−/<sup>−</sup> mice experienced perinatal death with variable phenotypes at late embryological stages and birth; these phenotypes included umbilical hernia, exencephaly, and a kinked tail, as well as complete loss of IdoUA residues in CS/DS chains [49]. However, a minority of embryos exhibited normal lung, bone, and cartilage features.

These findings indicate that DS, DS-PGs, DS-epi1 and/or DS-epi2 are important in early embryonic development and perinatal survival [49].

#### **6. Ongoing Projects and Future Perspectives**

Our group established a multicenter collaboration network to promote comprehensive basic research on mcEDS. In this network, genetic testing is provided as part of routine medical care covered by national health insurance, using a custom next generation sequencing-based panel that includes all EDS-related genes and genes for hereditary connective tissue disorders [50]. Whole exome sequencing is performed to identify other potential causative genes for mcEDS in patients with similar features who do not exhibit pathogenic variants in *CHST14* or *DSE*. Biochemical abnormalities of the extracellular matrix are investigated (e.g., various types of collagen and DS-PGs, including decorin and biglycan), using patient skin fibroblasts. Furthermore, mass spectrometry is performed to identify appropriate serum/plasma biomarker(s) that may be useful in the diagnosis or surveillance of disease progression. A subsequent pathological approach includes an elucidation of the structural alterations of collagen fibril networks and GAG chains of decorin in skin specimens from patients with mcEDS-*DSE*. Determination of the crystal structure of D4ST1 would be useful for understanding the effects of common missense variants (e.g., p.(Pro281Leu)) in *CHST14*. A technical breakthrough in the efficient generation of knockout mice for *Chst14* (e.g., CRISPR/cas9) is needed to continue mouse-based phenotypic and pathophysiological investigations. Considering the differences in the phenotypes between patients with mcEDS and *Chst14*−/<sup>−</sup> or *DS-epi1*−/<sup>−</sup> mice, experiments using other models that could reflect phenotypic and pathophysiological abnormalities in patients with mcEDS, such as induced pluripotent stem cells, would be particularly useful. All of these approaches will be valuable for further elucidating the critical roles of DS and DS-PGs, including decorin and biglycan, in the multisystem development and maintenance of connective tissues; in addition, they provide fundamental evidence for future etiology-based therapies, such as adeno-associated virus-based gene therapy.

**Acknowledgments:** We thank all the collaborators engaged in this project. We are also grateful to the patients and their families for participating in the studies described here. We thank Ryan Chastain-Gross, from Edanz Group (www.edanzediting.com/ac) for editing a draft of this manuscript. The drafting and revision of this review was supported by the Practical Research Project for Rare/Intractable Diseases #105 from the Japan Agency for Medical Research Development (AMED) (T.K., S.M., T.Y., N.M., and S.Y.); Grant-in-Aid for Young Scientists (B) (16K19396) (T.Y.); Grant-in-Aid for Scientific Research (C) from the Ministry of Education, Culture, Sports, Science and Technology of Japan (#16K08251, #19K07054) (S.M.) (#19K08745) (T.Y.); Grant-in-Aid for Scientific Research (B) from the Ministry of Education, Culture, Sports, Science and Technology of Japan (#19H03616) (T.K., S.M., T.W., T.Y., N.M., and S.Y.); Grant-in Aid for Research Center for Pathogenesis of Intractable Diseases from the Research Institute of Meijo University (S.M. and S.Y.).

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

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

### **Defining the Clinical, Molecular and Ultrastructural Characteristics in Occipital Horn Syndrome: Two New Cases and Review of the Literature**

**Aude Beyens 1,2, Kyaran Van Meensel 1, Lore Pottie 1, Riet De Rycke 3,4,5, Michiel De Bruyne 3,4,5, Femke Baeke 3,4,5, Piet Hoebeke 6, Frank Plasschaert 7, Bart Loeys 8, Sofie De Schepper 2, Sofie Symoens <sup>1</sup> and Bert Callewaert 1,\***


Received: 28 June 2019; Accepted: 11 July 2019; Published: 12 July 2019

**Abstract:** Occipital horn syndrome (OHS) is a rare connective tissue disorder caused by pathogenic variants in ATP7A, encoding a copper transporter. The main clinical features, including cutis laxa, bony exostoses, and bladder diverticula are attributed to a decreased activity of lysyl oxidase (LOX), a cupro-enzyme involved in collagen crosslinking. The absence of large case series and natural history studies precludes efficient diagnosis and management of OHS patients. This study describes the clinical and molecular characteristics of two new patients and 32 patients previously reported in the literature. We report on the need for long-term specialized care and follow-up, in which MR angiography, echocardiography and spirometry should be incorporated into standard follow-up guidelines for OHS patients, next to neurodevelopmental, orthopedic and urological follow-up. Furthermore, we report on ultrastructural abnormalities including increased collagen diameter, mild elastic fiber abnormalities and multiple autophagolysosomes reflecting the role of lysyl oxidase and defective ATP7A trafficking as pathomechanisms of OHS.

**Keywords:** occipital horn syndrome; Ehlers–Danlos syndrome type IX; Menkes syndrome; ATP7A; review; copper transport; elastic fiber; collagen

#### **1. Introduction**

Occipital horn syndrome (OHS, OMIM#304150), previously known as Ehlers–Danlos syndrome type IX or X-linked cutis laxa, is a rare disorder characterized by prominent connective tissue abnormalities including cutis laxa, hernias, joint laxity and bladder diverticula, and pathognomonic exostoses with "occipital horns" or downward pointing exostoses situated in the tendinous insertions of the sternocleidomastoid and trapezius muscles. In addition, patients may show mild to moderate intellectual disability [1–6]. Being allelic with Menkes disease (MD, OMIM#30011), OHS is considered the milder end of the phenotypic spectrum due to pathogenic variants in *ATP7A* [4] that encodes a copper transporter. Several clinical features in both disorders are related to the malfunctioning of cupro-enzymes, including lysyl oxidase, dopamine ß-hydroxylase, tyrosinase and cytochrome C oxidase [7–10].

ATP7A mediates (1) copper transport from the gastrointestinal tract into the bloodstream, (2) intracellular delivery of copper to cupro-enzymes in the secretory pathway and (3) efflux of excess copper from the cell [11,12]. Intracellular ATP7A trafficking is essential for copper homeostasis. Under basal low copper conditions, ATP7A is found in the trans-Golgi network, where it is essential for cupro-enzyme biogenesis. At higher copper concentrations, it reversibly localizes to the plasma membrane and post-Golgi compartments, where it is responsible for the extrusion of excess copper from the cell [13–15]. The *ATP7A* gene maps to Xq13.3 and spans a genomic region of ~140 kb. Its predominant transcription product contains 23 exons [16–20]. ATP7A belongs to the large P-type ATPase family, which are ATP-driven membrane pumps essential in the maintenance of electrochemical gradients (such as Na+/K+-, H+/K<sup>+</sup>- and Ca2<sup>+</sup>-pumps), as well as lipid and cationic homeostasis [21], including ATP7A and ATP7B, both important for copper homeostasis [13,22]. P-type ATPases share a structural core containing a transmembrane (TM) domain responsible for transport and an N-, P- and A- soluble catalytic domain, respectively involved in nucleotide binding, phosphorylation and dephosphorylation [23]. The transmembrane domain of ATP7A contains eight transmembrane helices (TM1–TM8). Six amino terminal domains contain metal-binding CXXC motifs (MBD1–MBD6) [24,25].

MD is mostly caused by truncating variants, including nonsense variants, frameshift variants and splice-site variants that lead to out of frame transcripts and large deletions, which result in very low to absent ATP7A levels [12,26]. On the contrary, the milder symptoms in OHS are the consequence of pathogenic variants in *ATP7A*, mostly "leaky" splice-site variants resulting in exon-skipping, which permit the production of a small amount of normal ATP7A mRNA [6,13,27–30]. This low amount of normal mRNA can result in residual protein activity and allow effective cation transfer to some copper-dependent enzymes. ATP7A levels as low as 2–5% have been shown to be sufficient to result in the milder OHS phenotype [31]. Connective tissue abnormalities in OHS have been attributed to a decreased activity of lysyl oxidase (LOX), a cupro-enzyme that normally deaminates lysine and hydrolysine residues in the first step of collagen crosslinking [7,32]. Patients may also encounter mild neurological signs and dysautonomia due to partial defects in dopamine-ß-hydroxylase enzyme activity, that converts dopamine to norepinephrine, a crucial neurotransmitter in norepinephrinergic neurons. As such, patients with OHS typically have low to normal levels of serum copper and ceruloplasmin together with abnormal plasma and cerebrospinal fluid (CSF) catecholamine levels [3,33,34].

Due to the rarity of OHS and the limited number of cases scattered in literature, the natural history and clinical phenotype remain incompletely studied. This study reports the clinical, molecular and ultrastructural data for two new cases of OHS. In addition, we performed a clinical review of 32 previously described patients [6,7,11,27–31,35–51].

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

#### *2.1. Patients*

Informed consent was obtained from subject 1 and his parents, and from subject 2. Additional consent was given to publish the clinical pictures in Figure 1. Clinical information was obtained by evaluating both patients at our center with the use of a detailed clinical checklist. We performed a literature review of 32 OHS patients with confirmed *ATP7A* pathogenic variants in the Pubmed and Embase databases, only including studies with ample clinical data. Selection of the reports was undertaken independently by two different authors (A.B and K.V.M). Clinical data from the literature were evaluated using the same methodology (Supplementary Table S1). This study was conducted in accordance with the Declaration of Helsinki and approved by the Ghent University Hospital ethics committee (registration number B670201319316).

#### 2.1.1. Clinical Reports

Case 1

Subject 1 (F1:II-1) was born at term after an uncomplicated pregnancy. At birth, his weight was 2400 g, length 45.5 cm and head circumference 32.4 cm. Developmental milestones were achieved normally. At three years of age, he presented with micturition problems evolving to urinary retention, which led to the discovery of giant bladder diverticula. Over subsequent years, several diverticulectomies were performed, followed by continent diversion and intermittent catheterization.

Clinical examination at 11 years of age showed mild facial dysmorphism (Figure 1) with a flat face, deep-set eyes, a long narrow nose, large ears and mild webbing of the neck. His weight was 24.9 kg (P1, −2.37 SD), height 139.2 cm (P26, −0.649 SD) and head circumference 56.2 cm (P98, +2.1 SD). Skeletal abnormalities included joint hyperlaxity of the wrists, metacarpophalangeal joints and fingers, a flat back, marked bony thickening of the proximal radius and tibia, genua valga and pedes plani. There was skin webbing between the second and third toe. His skin was soft and mildly hyperextensible with fine wrinkles over the abdomen and dorsum of hands and feet. He had no hair abnormalities. Neurological examination was normal. He had recurrent inguinal hernias. He reported bruising easily, but no impaired wound healing. He suffered from chronic joint pain. The patient attended normal education.

Additional examinations included a total skeletal radiography and MRI/MRA (Figure 2). X-rays show severe skeletal dysplasia, occipital horns, undertubulation of bony structures and broadening of the ventral end of the first left rib, the distal end of both claviculae and the scapular neck. The long bones showed bowing with mid-diaphyseal broadening. There was bilateral luxation of the radial heads, bilateral overgrowth of the ulnar coronoid processes and prominent trochanter minor of both femurs. We observed a short fibula and bilateral coxa valga with rounded iliac wings. The metatarsals were broad. The vertebrae showed platyspondyly. Brain MRI and angiography showed tortuous intracranial arteries. Echocardiography, lung radiography and spirometry were normal. Serum copper and ceruloplasmin levels were within normal ranges (89.6μg/dL (70–140) and 0.287 g/L (0.2–0.6), respectively).

**Figure 1.** Clinical characteristics in subjects F1:II-1 and F2:II-1. Subject F1:II-1 displays mild facial dysmorphism with a flat face, deep-set eyes, a long and narrow nose, large ears and mild webbing of the neck. Craniofacial features in patient F2:II-1 are more pronounced and include dolichocephaly, a high forehead, biparietal narrowing, downslanting eyes, a convex nasal ridge, midfacial hypoplasia, micrognathia and low-set ears. No hair abnormalities were observed. Both have lax and hyperextensible skin, marked joint hypermobility and deformities of the proximal radius.

**Figure 2.** Total skeletal radiography and MRI/MRA studies in subject F1:II-1 showing severe skeletal dysplasia with occipital horns, undertubulation of bony structures and broadening of the ventral end of the first left rib, both claviculae and scapular neck, bowing of the long bones with mid-diaphyseal broadening, bilateral luxation of the radial heads, bilateral overgrowth of the ulnar coronoid processes, prominent trochanter minor of both femurs, short fibula, bilateral coxa valga with rounded iliac wings and broad metatarsals. Brain MRI and angiography showed tortuous intracranial arteries and normal brain structure.

#### Case 2

Subject 2 (F2:II-1) was evaluated at the age of 30. He was born at term after an uncomplicated pregnancy. He developed an inguinal hernia at six weeks of age. Since then, he had multiple inguinal and femoral hernias requiring repeated mesh repairs. Starting from two years of age, he had severe bladder problems, including recurrent urinary tract infections and urinary retention, due to multiple bladder diverticula. After more than 20 surgical bladder repairs, he started self-catheterization at 8 years of age. Aged 20 years, he underwent left nephrectomy for a non-functioning kidney. He reported a mild motor delay and walked at two years of age. Joint hypermobility resulted in repeated shoulder and elbow subluxations. He was of normal intelligence and pursued higher education. He suffered from migraine, worsening musculoskeletal pain, fatigue, and depression for which he was treated with gabapentin and amitriptyline. He had a history of mitral valve prolapse, orthostatic hypotension, palpitations and a prolonged QT interval on electrocardiogram. Finally, he had a longstanding history of upper and lower gastrointestinal symptoms including nausea, post-prandial vomiting and chronic diarrhea, possibly in the context of dysautonomia. He was diagnosed with asthma.

His height was 169 cm, his weight was 50.6 kg and his head circumference was 58.5 cm. Craniofacial features consisted of dolichocephaly, a high forehead, biparietal narrowing, downslanting eyes, a convex nasal ridge, midfacial hypoplasia, micrognathia and low-set ears. He had lax and hyperextensible skin (Figure 1). He had no hair abnormalities. Skeletal abnormalities included occipital horns, a loss of sagittal spine curvature, scoliosis, pectus excavatum, and restricted elbow extension. He had bowing of the ulna and radius, bilateral exostoses of the radii and tibiae, bilateral rounding of the iliac wings, cranial displacement of the right humeral head and genua vara. Neurological examination and brain

MRI were normal. Serum copper and ceruloplasmin levels were 8.3 μmol/L (11–20) and 0.13 g/L (0.20–0.50), respectively.

#### *2.2. Molecular Analysis*

For individual 1, genomic DNA was extracted from blood leukocytes using standard procedures. Array comparative genomic hybridization (180 k arrayCGH, Agilent Technologies, Santa Clara, CA, USA) showed a 300 kb deletion of chromosome band 1q31.1 (1q31.1.1q31.1(188407205-188707983)x1)(hg19/GRCh37), which does not contain any known genes. *FBLN5* (OMIM#219100, GenBank RefSeq accession number NM\_006329.3), *LTBP4* (OMIM#613177, NM\_003573.2) and *ATP7A* (OMIM#304150, NM\_000052.7) gene sequencing was performed without the identification of a pathogenic mutation. Total RNA was extracted from cultured fibroblasts using the RNeasy Kit (Qiagen, Hilden, Germany) and cDNA was synthesized with the iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA, USA). *ATP7A* cDNA analysis was performed followed by confirmation with long-range RT-PCR and next-generation sequencing (Miseq, Illumina, San Diego, CA, USA). For individual 2, array comparative genome hybridization was normal after which *ATP7A* gene sequencing (GenBank RefSeq accession number NM\_000052.7) was performed.

#### *2.3. Transmission Electron Microscopy (TEM)*

For transmission electron microscopy, 3 mm skin fragments from both individuals and an age- and sex-matched control were immersed in a fixative solution of 2.5% glutaraldehyde and 4% formaldehyde in Na-Cacodylate buffer 0.1 M, placed in a vacuum oven for 30 min and left rotating for 3 hours at room temperature. This solution was later replaced with fresh fixative and samples were left rotating over night at 4 ◦C. After washing, samples were post-fixed in 1% OsO4 with K3Fe(CN)6 in 0.1 M NaCacodylate buffer, pH 7.2. After washing in double-distilled H2O, samples were subsequently dehydrated through a graded ethanol series, including bulk staining with 2% uranyl acetate at the 50% ethanol step, followed by embedding in Spurr's resin. To select the area of interest on the block and in order to have an overview of the phenotype, semi-thin sections were first cut at 0.5 μm and stained with toluidine blue. Ultrathin sections of a gold interference color were cut using an ultramicrotome (Leica EM UC6, Wetzlar, Germany), followed by post-staining in a Leica EM AC20 for 40 min in uranyl acetate at 20 ◦C and for 10 min in lead stain at 20 ◦C. Sections were collected on formvar-coated copper slot grids. Grids were viewed with a JEM 1400plus transmission electron microscope (JEOL, Tokyo, Japan) operating at 80 kV.

#### *2.4. Collagen Biochemical Analysis*

Fibroblast cultures were started from skin biopsies from individuals 1 and 2, and a matched control. Biochemical (pro)collagen sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE analysis) was performed on the medium and cellular fractions of the cultured skin fibroblasts. When confluent, cells were labeled with 14C-proline [52]. After SDS-PAGE separation, further processing of the gels was performed as described previously by our group [53].

#### **3. Results**

#### *3.1. Clinical Characteristics*

We describe the clinical characteristics of two new individuals with OHS and 32 patients described in the literature (Table 1, Supplementary Table S1). All patients were male, except for patient F17:II-2. The mean reported age of patients was 19.6 years (mean 17, range 1–50). Disease-related mortality was observed in 5 out of 34 patients. The mean age at death was 21.2 years and causes included respiratory failure (post-surgery apnea), gastric perforation and diffuse intravascular coagulation due to a large cephalhematoma at birth.


**Table 1.** Overview of the phenotypic presentation in occipital horn syndrome.

Initial presentations leading to the diagnosis are presented in Table 2. One-third of patients presented with neurological features including hypotonia, developmental delay, and seizures. Nine individuals presented with connective tissue problems such as cephalhematoma and inguinal hernia. Other presenting symptoms included bladder problems such as bladder diverticula, urinary tract infections and skeletal problems.


**Table 2.** Initial presentations leading to the diagnosis in occipital horn syndrome.

"CTD": connective tissue disease.

Patients displayed variable craniofacial features, including a long face (6/13, 46%), large ears (5/13, 38%), sagging cheeks (5/11, 45%) and hair abnormalities (14/19, 74%). Trichoscopy confirmed pili torti in 7 out of 10 examined individuals. Connective tissue abnormalities were reported in all subjects. Cutis laxa was present in the majority of patients (25/26, 94%), as well as inguinal hernias (15/25, 63%). An umbilical hernia and hiatal hernia were observed in only one patient each (F25:II-1 and F15:II-1, respectively). Wound healing was reported to be normal after trauma or surgery. Occipital horns were reported in nearly all cases (25/26, 96%). Other recurrent skeletal manifestations included hammer-shaped claviculae (9/19, 47%), scoliosis (8/21, 38%), pectus deformities (13/23, 57%), radial/tibial exostoses (6/16, 38%), coxa valga (6/16, 38%) and genua valga (6/17, 35%). Dislocations, mainly of the radial heads and joint hypermobility were present in about two-thirds of individuals, 12/20 (60%) and 16/26 (62%), respectively. Less frequently reported skeletal manifestations include bowing of the long bones, mid-diaphyseal broadening, metaphyseal spurring and rounding of the iliac wings. Osteopenia or osteoporosis was reported in four patients and fractures were observed in two subjects.

Mild, mainly motor, developmental delay, was observed in 17 subjects (17/30, 57%). Intellectual disability was present in an equal number of patients (17/33, 52%) of whom seven had mild, four had moderate and three had severe intellectual disability. In three patients, the severity of intellectual disability was not specified. Several subjects displayed muscle hypotonia, especially at a young age (before the age of two). Seizures were documented in five patients (5/25, 20%). One individual was diagnosed with a spontaneous intracerebral hemorrhage at the age of 15, resulting in aphasia and spastic quadriparesis. MRI brain studies were performed in 11 patients and documented to be abnormal in 7 subjects, showing mild atrophy and/or delayed myelination. Intracranial tortuosity was present in two-thirds of patients (7/11, 64%). Tortuosity of extracranial arteries was found in three cases and was located in the cervical, splenic and hepatic arteries. Aneurysm formation was found in five patients, mainly affecting the same arteries. One individual was diagnosed with a type A aortic dissection at the age of 39, which was successfully surgically manage. Two patients showed dilatation of the large veins. Orthostatic hypotension, temperature instability, chronic diarrhea, and other symptoms of dysautonomia were seen in a high number of patients (13/15, 87%). Severe urological complications are an important concern in OHS, with bladder diverticula in 83% (25/30) of reported patients, of which

three were eventually diagnosed with a bladder rupture. Recurrent urinary tract infections based on incomplete emptying of the bladder were seen in 74% (17/23) of patients. One-third of patients had vesicourethral reflux (VUR), sometimes complicated by secondary renal problems. Asthma was reported in four cases. Low levels of serum copper and ceruloplasmin were found in two-thirds of cases, 20/28 (71%) and 20/27 (74%), respectively.

#### *3.2. Molecular Characteristics*

After normal *ATP7A* gene sequencing, cDNA analysis showed a deletion of exon 10 in subject F1:II-1. Subsequent confirmation with long-range RT-PCR revealed a deep intronic c.2407-516G>T variant, underlying the exon 10 skipping, likely due to the activation of a cryptic splice acceptor. In patient F2:II-1, *ATP7A* gene sequencing identified a hemizygous c.2917-4A>G splice-site variant in *ATP7A*.

A total of 16 different *ATP7A* pathogenic variants were reported (Figure 3); these included seven splice-site, four missense and two frameshift variants, two deep intronic variants and one deletion. In eight patients, the specific variant was not reported, including a 98-bp deletion in the regulatory region (F9:II-1), a deep intronic intron six splice donor site variant (F12:II-1) and a duplication in exon 11 and 12 (F23:II-1). Nonsense ATP7A variants were not observed. Pathogenicity information and CADD scores can be found in Supplementary Table S2. We did not observe any clear genotype-phenotype correlation and there was no correlation with serum copper and ceruloplasmin levels (Supplementary Table S3).

**Figure 3.** ATP7A pathogenic variants reported in this study. ATP7A consists of six amino terminal metal binding domains (MBD1-MBD6), a transmembrane domain containing eight transmembrane helices (TM1-TM8) and an N-, P- and A- soluble catalytic domain. The pathogenic variants identified in subjects F1:II-1 and F2:II-1 are indicated in bold and outlined.

#### *3.3. Transmission Electron Microscopy*

Transmission electron microscopy (TEM) findings in skin samples of subjects F1:II-1, F2:II-1 and a matched control are shown in Figure 4. Collagen fibrils in the dermis of the control sample are regularly organized in curvilinear bundles with uniform diameters. In the dermis of both OHS patients, the collagen fiber diameter appears to be larger and more variable. The fibrils are more loosely packed (especially in F2:II-1) but retain a normal shape. TEM of a control elastic fiber demonstrates a solid and dense core of elastin surrounded by a sparse mantle of microfibrils. In both individuals, the elastin core is surrounded by a debris extending outside the elastic fiber with loss of the microfibrillar organization. In addition, the elastin core in subject F2:II-1 is almost completely disrupted and difficult to discriminate. Fibroblast abnormalities are observed in both patients, with multiple autophagolysomes containing sickle-shaped electron dense deposits (Supplementary Figure S1). The mitochondria, Golgi complex, and endoplasmatic reticulum appear normal.

**Figure 4.** Transmission electron microscopy findings in occipital horn syndrome. Transmission electron microscopy (TEM) findings in skin samples of patients F1:II-1 and F2:II-1 and a matched control showing collagen (column 1), elastic fibers (column 2) and a fibroblast (column 3). col = collagen; ef = elastic fiber; fb = fibroblast. Magnification for column 1: ×15,000, column 2: ×8000 and column 3: ×4000.

#### *3.4. Collagen Biochemistry*

Considering the collagen abnormalities observed in both individuals with transmission electron microscopy, SDS-PAGE analysis of the intracellular and secreted (pro)collagen proteins produced by skin fibroblasts was performed for subjects 1 and 2 and a matched control. In both cases, we observed normal SDS-PAGE migration patterns (Supplementary Figure S2).

#### **4. Discussion and Conclusions**

The absence of large case series and natural history studies precludes efficient diagnosis and management of OHS patients. This study describes the clinical and molecular characteristics of two new patients and 32 patients previously reported in the literature. OHS is considered a mild but rare allelic variant of Menkes disease. OHS patients show prominent connective tissue abnormalities and have a better prognosis and survival rate than patients with MD [3], with low childhood mortality (1/34) in OHS. However, there is still significant mortality in (young) adults, often related to gastrointestinal, respiratory, or bleeding complications, such as an intestinal perforation due to a gastric ulcer and (post-surgery) apnea.

The clinical characteristics of OHS are listed in Table 2. Craniofacial features include mild facial dysmorphism with a long face, large ears and sagging cheeks in about half of reported patients. Hair abnormalities, including coarse hair and pili torti, are frequently present. Ubiquitous connective tissue manifestations in OHS include increased skin laxity and inguinal hernia.

Concerning the skeletal features, the pathognomonic occipital horns are present in all patients but one, which is probably an age-dependent discrepancy [41,54,55]. Other less specific skeletal abnormalities include hammer-shaped claviculae and radial/tibial exostoses. Scoliosis, pectus deformities, coxa and genua valga, joint luxations, mainly of the radial heads and general joint hypermobility are frequent but non-specific. These skeletal problems may limit daily activities and often require multiple surgical interventions.

Urological complications, mostly giant bladder diverticula and vesicourethal reflux, are found in over 80% of subjects. Secondary problems such urinary tract infections, urinary retention, bladder rupture, or VUR-mediated renal failure, which are all related to the bladder diverticula, are of great concern. Surgical interventions removing the diverticula often fail, with frequent relapse of the diverticula. Self-catheterization, vesicostomy, or continent diversion are often required for proper emptying of the bladder.

Previous reports describe a mild neurological phenotype limited to slight generalized muscle weakness and symptoms of dysautonomia [3,27], however, more detailed case studies reveal some conspicuous observations including developmental delay and intellectual disability in two-thirds of subjects and documented seizures in five individuals. Nevertheless, neurological affliction remains rather mild and normal intelligence is possible. Symptoms of dysautonomia, including chronic diarrhea, temperature instability, and orthostatic hypotension are present in almost 90% of patients and can be disabling.

Brain imaging studies are rarely performed in OHS patients and are often reported as normal. However, intracranial arterial tortuosity are noted in approximately two thirds of patients. Extracranial arterial tortuosity is less frequently observed, but has been reported in the cervical, splenic and hepatic vasculature. Aneurysm formation may complicate pre-existing tortuosity and may affect the arterial and venous circulation.

OHS is often diagnosed upon the identification of connective tissue anomalies, including cutis laxa, joint hypermobility with hypotonia, mild intellectual impairment and bladder diverticula. Later on, the pathognomonic occipital horns become more evident. The initial differential diagnosis includes other forms of cutis laxa, including *FBLN5*- and *LTBP4*-related cutis laxa (OMIM#219100 and #613177) [56] and the dermatosparaxis type of Ehlers–Danlos syndrome (EDS type VIIC, OMIM#225410) [57], which may present with cutis laxa, hypotonia and bladder diverticula, but with normal mental development. However, in OHS there is no early onset emphysema as in *FBLN5*- and *LTBP4*- related cutis laxa, nor easy bruising or severe skin fragility as in the dermatosparaxis type of EDS. *ATP6V0A2*-related cutis laxa (OMIM#219200) may also present with mild intellectual impairment but does not commonly show bladder diverticula or exostoses. Later on the exostoses can be mistaken for hereditary multiple exostoses due to defects in *EXT1* or *EXT2* (OMIM#133700 and #133701) [58].

Concerning follow-up, routine evaluation of neurodevelopment and early intervention with physiotherapy is recommended, as well as regular urological evaluation, including video urodynamic studies of bladder function. Any pain, or vascular or neurological symptoms that may relate to exostoses should be promptly investigated. To date, there are no data on a possible risk for chondrosarcomas in exostoses although these have been described in hereditary multiple exostoses [59]. It could be recommended to perform MR angiography and echocardiography from puberty onwards. Indeed, aortic root dilatation and dissection, and an increased risk for intracranial bleeds and ischemic complications have been described in other related connective tissue disorders such as Ehlers–Danlos, arterial tortuosity, Marfan and some cutis laxa syndromes. Furthermore, spirometry might detect asthma at an early stage as it was reported in four individuals. Prolonged vigilance for apnea is necessary following surgery.

The molecular data confirmed that pathogenic variants in *ATP7A* causing OHS mainly included splice-site variants, while missense and frameshift variants were less frequently observed. This observation may fit the hypothesis that a low amount of normal *ATP7A* mRNA and residual ATP7A functioning is sufficient to prevent the more severe MD [13,27,29,31,60]. No mutational hot spots were observed (Figure 3). Importantly, nearly all pathogenic variants are located in the endofacial domains and only two affect a transmembrane domain or cytosolic loop, respectively. This highlights the critical role of the transmembrane domains forming the channel for copper translocation [13]. Variants in OHS mainly affect residues implicated in ATPase-coupled copper transport. Some of these variants are predicted to inhibit the catalytic dephosphorylation leading to permanent distribution of the protein to the plasma membrane and cytosolic vesicles [14,15,21,61,62].

Our transmission electron microscopy data in skin samples of OHS individuals show larger diameters of collagen fibrils that were more loosely packed compared to those of the control sample. Elastic fibers were affected to a lesser extent but are fragmented and/or surrounded by extracellular matrix debris. OHS is associated with decreased activity of the copper dependent lysyl oxidase (LOX), that oxidates lysyl and hydroxylysyl residues in collagen and lysyl residues in elastin to aldehydes [7]. These aldehyde residues are than condensed to form cross-links, essential for collagen fibril stabilization and the integrity and elasticity of mature elastin [32,63]. Overall, this would not affect the amount or migration of soluble collagens from the medium or cellular fraction upon electrophoresis but will affect the assembly in the extracellular matrix. Also, lysyl oxidase may form cross-links with the N-propeptide domain of collagen type V, essential in collagen diameter regulation [64,65]. Fibroblasts synthesize only low amounts of type V collagen which is difficult to evaluate by electrophoresis [66,67]. Indeed, irregular fibril diameters are reminiscent of classic Ehlers–Danlos syndrome, despite the absence of cauliflower fibrils [68].

The observation of major fibroblast abnormalities, with multiple autophagolysosomes containing sickle-shaped electron dense deposits, is novel and might point to defective trafficking and trapping of proteins needing cuproenzyme-dependent processing in the post-Golgi network [14,15,21].

In conclusion, this study sheds light on the natural history of OHS by describing two new patients and 32 patients reported in the literature. Mortality in adulthood warrants long-term specialized care and follow-up, incorporating MR angiography, echocardiograpy and spirometry into standard follow-up guidelines for OHS patients, next to orthopedic and urological follow-up. Ultrastructural abnormalities including increased collagen diameter, mild elastic fiber abnormalities, and multiple autophagolysosomes reflect the role of LOX and defective ATP7A trafficking in the pathomechanism of OHS.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/7/528/s1; Supplemental Figure S1: Fibroblast abnormalities in OHS. Magnification: x4000, x8000 and x15000; Supplemental Figure S2: SDS-PAGE collagen analysis in patients F1:II-1 and F2:II-1 and a matched control. Panel A: medium procollagen fraction on fibroblasts in culture for three days. Panel B: cellular collagen fraction; Supplemental Table S1. Clinical characteristics of all patients. '"+": present, "-": absent, "M": male, "F": female, "y": year(s), "mo": month(s), "d":day(s), "CVA": cerebrovascular accident, "DIC": diffuse intravascular coagulation, "FTT": failure to thrive, "GER": gastro-esophageal reflux, "SC": subcutaneous, "UTI": urinary tract infection; Supplemental Table S2. Pathogenicity information for the ATP7A variants in this study. The pathogenicity classification used in this table is based on the Alamut®mutation analysis software, which uses the following prediction tools: Align GVGD (Huntsman Cancer Institute, University of Utah) combines the biophysical characteristics of amino acids and protein multiple sequence alignments to where missense substitutions in genes of interest fall from enriched deleterious to enriched neutral. SIFT predicts whether amino acid substitution affects protein function. The MutationTaster prediction program integrates information from different biomedical databases, evolutionary conservation, splice-site changes and loss of protein features. Protein conservation is based on MARRVEL®multiple protein alignment. The splice prediction predicts the chance of variants having

an impact on alternative splicing. Gnomad allele frequencies and Clinvar data can also be found. In the last column, CADD scores are depicted (Kircher et al., 2014). "NS": not specified, "NA": not applicable; Supplemental Table S3. Copper and ceruloplasmin levels in all patients.

**Author Contributions:** Conceptualization, A.B. and B.C.; Data curation, A.B., K.V.M., P.H., F.P., B.L., S.S. and B.C.; Formal analysis, K.V.M.; Funding acquisition, B.C.; Investigation, A.B., K.V.M., L.P., R.D.R., M.D.B., F.B., S.S. and B.C.; Methodology, A.B., L.P., R.D.R., S.S. and B.C.; Project administration, B.C.; Resources, K.V.M., R.D.R., P.H., F.P., B.L., S.S. and B.C.; Supervision, S.D.S., S.S. and B.C.; Validation, A.B., K.V.M. and B.C.; Visualization, A.B., L.P. and R.D.R.; Writing—original draft, A.B.; Writing—review and editing, A.B., K.V.M., L.P., R.D.R., M.D.B., F.B., P.H., F.P., B.L., S.D.S., S.S. and B.C.

**Funding:** This work is supported by the Special Research Fund, Flanders of Ghent University (grant 01N04516C to B.C.) and the European Academy of Dermatovenereology (EADV project PPRC-2018-50 to B.C.). B.C. and B.L. are senior clinical investigators of the Research Foundation Flanders.

**Acknowledgments:** We are thankful to all the families and patients for their interest and participation. We want to thank Zoë Malfait and Lisa Caboor for their technical support. The Center for Medical Genetics and Department of Dermatology of the Ghent University Hospital are members of the European Reference Network (ERN)-Skin. The Centers for Medical Genetics of both Ghent and Antwerp are members of VASCERN.

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

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Characterization of Two Novel Intronic Variants <sup>A</sup>**ff**ecting** *Splicing* **in** *FBN1***-Related Disorders**

### **Carmela Fusco 1,\*, Silvia Morlino 2, Lucia Micale 1, Alessandro Ferraris 2, Paola Grammatico <sup>2</sup> and Marco Castori <sup>1</sup>**


Received: 19 April 2019; Accepted: 7 June 2019; Published: 10 June 2019

**Abstract:** *FBN1* encodes fibrillin 1, a key structural component of the extracellular matrix, and its variants are associated with a wide range of hereditary connective tissues disorders, such as Marfan syndrome (MFS) and mitral valve–aorta–skeleton–skin (MASS) syndrome. Interpretations of the genomic data and possible genotype–phenotype correlations in *FBN1* are complicated by the high rate of intronic variants of unknown significance. Here, we report two unrelated individuals with the *FBN1* deep intronic variants c.6872-24T>A and c.7571-12T>A, clinically associated with MFS and MASS syndrome, respectively. The individual carrying the c.6872-24T>A variant is positive for aortic disease. Both individuals lacked ectopia lentis. In silico analysis and subsequent mRNA study by RT-PCR demonstrated the effect of the identified variant on the splicing process in both cases. The c.6872-24T>A and c.7571-12T>A variants generate the retention of intronic nucleotides and lead to the introduction of a premature stop codon. This study enlarges the mutation spectrum of *FBN1* and points out the importance of intronic sequence analysis and the need for integrative functional studies in *FBN1* diagnostics.

**Keywords:** fibrillin 1; Marfan syndrome; MASS syndrome; mRNA; splicing

#### **1. Introduction**

The *FBN1* gene is located in 15q21.1 and includes 65 exons which encode for fibrillin 1, a large glycoprotein constituted by 47 cysteine-rich epidermal growth factor (EGF)-like domains and seven motifs homologous to the binding protein for transforming growth factor beta (TGFβ) [1,2]. Fibrillin 1 is a major structural component of the extracellular matrix microfibrils and thus, gives stability and elasticity to many tissues [2]. Variants in *FBN1* cause a wide range of autosomal dominant heritable connective tissue disorders, including Marfan syndrome (MFS; OMIM 154700), mitral valve–aorta–skeleton–skin syndrome (MASS syndrome; OMIM 604308), Marfan lipodystrophy syndrome (OMIM 616914), isolated autosomal dominant ectopia lentis (OMIM 129600), Weill–Marchesani syndrome type 2 (OMIM 608328), acromicric dysplasia (OMIM 102370), geleophysic dysplasia type 2 (OMIM 614185), stiff skin syndrome (OMIM 184900), and autosomal dominant thoracic aortic aneurysms and dissections [3].

To date, over 1800 different *FBN1* germline variants have been identified so far (UMD-*FBN1*, http://www.umd.be/FBN1/) [4,5] and only a few genotype–phenotype correlations are available [6,7]. For example, nonsense, frameshift, and some splicing variants appear correlated with a more severe skin and skeletal phenotype, as compared to in-frame variants [8]. Other studies suggest that variants affecting or creating cysteine residues are more commonly associated with ectopia lentis, while *null*

alleles typically combine with an increased rate of aortic events in young age and thoracic deformities [7]. In the suspicion of a *FBN1*-related disorder, the identification of the causative variant is not only relevant for diagnosis confirmation and genetic counseling but is also increasingly useful for personalized medicine approaches, comprising preventive drug treatment and aortic surgery planning [9].

The introduction of next generation sequencing (NGS) technologies in molecular diagnostics improved turnaround time and costs but did not significantly affect the clinical interpretation of variants of uncertain significance, such as intronic nucleotide changes which potentially affect splicing. The burden associated with uncertain results related to these variants is still high in MFS and related disorders, as point variants possibly impacting FBN1 pre-mRNA splicing seem to account for ~10% of reported molecular findings in MFS [10]. In these cases, optimal clinical interpretation should integrate molecular findings with customized studies exploring variant effects at the transcriptional or translational levels.

Here, we report two unrelated individuals with MFS and MASS syndrome, respectively, which were associated with different intronic variants in *FBN1*. Characterization of the effects of these variants on mRNA allowed us to support their pathogenicity and, therefore, to use them for patient and family management.

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

All investigated subjects signed informed consent for molecular testing and research use of biological and related clinical data. Molecular testing on cDNA and mRNA in individual 1 and 2 were carried out within the routine clinical diagnostic activities of the Division of Medical Genetics of Foundation IRCCS-Casa Sollievo della Sofferenza, San Giovanni Rotondo (Italy). The results of this work are part of a larger research project approved by our local ethics committee (approval protocol no. GTB12001).

#### *2.1. Clinical Description: Individual 1*

This was a 27-year-old man, the second child of unrelated parents. His father was 50-year-old, 191 cm tall and affected by dilatation of the aortic root and ascending aorta, which required surgery at 45 years. The mother and sister were healthy, and 172 and 161 cm tall, respectively. The proband was born at term after an uneventful pregnancy. At birth, he presented respiratory distress, which needed intensive care support for a few days. The neonatal period was otherwise normal, as well as psychomotor development, scholarship, and mentation. At 11 years, progressive scoliosis required an orthopedic corset for some years. The patient suffered from recurrent spontaneous pneumothorax with four episodes between 12 and 14 years of age. For this, he successfully underwent pleurodesis at 14 years. At 18 years, the previous diagnosis of aortic disease in the father prompted full cardiologic assessment in the proband who showed dilatation of the aortic root and mitral valve prolapse. Since then, he was under preventive pharmacologic therapy. Current treatment includes losartan 50 mg/day (single dose) and nebivolol 50 mg/day (single dose), without significant side effects. The last heart ultrasound, at 27 years, showed a stable dilatation of the aortic root (48 mm; Z score 171.13 = 4.66 SD [11]) with normal ascending aorta (34 mm), minimal insufficiency of the aortic valve, prolapse of the anterior mitral leaflet with minimal insufficiency, and mild insufficiency of the tricuspid valve. Dilatation of the aortic root (48 × 43 mm) was also confirmed by angio-computer tomography. A total spine MRI at 23 years showed bilateral dural ectasias of the lumbar spine. A recent ophthalmologic exam excluded lens dislocation and revealed mild myopia (−2 diopters on the right and −1.5 diopters on the left).

Physical examination included height 196 cm, weight 72 kg, arm span 198 cm, arm span/height ratio 1.01 (normal), bilateral positive wrist sign, bilateral negative thumb sign, apparent enophthalmos, down slanting palpebral fissures, malar hypoplasia, high-arched palate, severe scoliosis, pectus carinatum, bilateral valgus deformity of the elbow, bilateral metatarsus varus, some mildly atrophic

post-surgical scars of the thorax, and striae rubrae of the back (Figure 1A–D). The combination of aortic root dilatation (Z-score > 2 SD) and 10 points of systemic features lead to the clinical suspect of MFS.

**Figure 1.** Clinical features. Individual 1 showing high-arched palate (**A**), pectus carinatum (**B**), scoliosis (**C**) and arachnodactyly (**D**). Lumbar spine MRI of Individual 2 demonstrating bilateral dural ectasias at axial (**E**), coronal (**F**), and sagittal (**G**) views. Asterisks indicate the dilated dural sac.

#### *2.2. Clinical Description: Individual 2*

Individual 2 was a 48-year-old, nullipara woman, the seventh child of a sibship of eleven. All brothers and sisters were healthy, but two brothers were described with tall stature (~194 cm). The mother was 165 cm tall and died at 65 years due to rupture of a cerebral aneurysm. The father was 175 tall and died at 78 years due to a stroke. Family history was negative for aortic disease. The proband was sent to clinical genetics assessment by the neurosurgeon, who first evaluated the woman for chronic back pain, diffuse spondylosis, and marked dural ectasias of the lumbar spine (Figure 1E–G). Relevant additional clinical features included moderately severe scoliosis since her teens, severe myopia (−6 diopters on the right and −8 dipters on the left), and chronic fatigue. A heart ultrasound repeatedly showed normal aortic diameters, prolapse, and mild insufficiency of the mitral valve. At 47 years, aortic root diameter was 36 mm (Z score 1.26 SD [11]). Routine ophthalmologic assessments always excluded lens dislocations. Physical examination showed height 181 cm, weight 80 kg, arm span 188 cm, arm span/height ratio 1.038 (normal), bilaterally positive wrist and thumb signs, hypermobility of fingers (but the Beighton score was 0/9), elbow limitation on both sides, enophthalmos, down slanting palpebral fissures, retrognathia, high-arched palate, striae distensae

of the shoulders. The presence of 10 points of systemic features and an aortic root diameter <2 SD suggested a diagnosis of the MASS syndrome. Positive family history for tall stature and absence of aortic disease in relatives were in accordance with the diagnosis. Molecular testing in Individual 2 was carried out with a multigene panel approach including extended intronic regions around exon–intron junctions (see below).

#### *2.3. Genomic DNA Extraction*

Genomic DNA was extracted from individuals' peripheral bloods by using Bio Robot EZ1 (Qiagen, Hilden, Germany), according to standard procedures. The DNA was quantified with Nanodrop 2000 C spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

#### *2.4. Genetic Testing (Sanger Sequencing): Individual 1*

Molecular analysis for individual 1 was performed by Sanger sequencing. Primers were designed using the Primer 3 Output program (http://frodo.wi.mit.edu/primer3/) to amplify extended intronic sequences and exon–intron flanking sequences of *FBN1* (RefSeq NM\_000138). Primers were checked both by BLAST and BLAT against the human genome to ensure specificity. Amplified products were subsequently purified and sequenced with a ready reaction kit (BigDye Terminator v1.1 Cycle, Applied Biosystems, Foster City, CA, USA). Fragments were then purified using DyeEx plates (Qiagen, Hilden, Germany) and resolved on an automated sequencer (3130xl Genetyc analyzer DNA Analyzer, ABI Prism, Foster City, CA, USA). Sequences were analyzed using the Sequencer software (Gene Codes, Ann Arbor, MI, USA). To investigate familial segregation, both parents were sequenced for the identified variant.

#### *2.5. Genetic Testing (Next-Generation Sequencing): Individual 2*

Individual 2's DNA underwent sequencing with a HaloPlex gene panel (Agilent Technologies, Santa Clara, CA, USA) designed to selectively capture known genes associated with syndromic and non-syndromic thoracic aneurysms and/or Mafanoid habitus, including: *ACTA2* (NM\_001141945), *BGN* (NM\_001711.5), *CBS* (NM\_000071.2), *COL3A1* (NM\_000090), *COL4A1* (NM\_001303110), *COL5A1* (NM\_000093), *COL5A2* (NM\_000393), *DLG4* (NM\_001365.4), *EFEMP2* (NM\_016938), *ELN* (NM\_000501), *EMILIN* (NM\_007046.3), *FBN1* (NM\_000138), *FBN2* (NM\_001999), *FLNA* (NM\_001110556), *FOXE3* (NM\_012186), *GATA5* (NM\_080473), *LOX* (NM\_002317), *MAT2A* (NM\_005911), *LTBP3* (NM\_001130144.2), *MAT2A* (NM\_005911.5), *MED12* (NM\_005120.2), *MFAP5* (NM\_003480), *MYH11* (NM\_001040113), *MYLK* (NM\_053025), *NOTCH1* (NM\_017617), *PLOD1* (NM\_001316320), *PRKG1* (NM\_006258), *SKI* (NM\_003036), *SLC2A10* (NM\_030777), *SMAD2* (NM\_005901.5), *SMAD3* (NM\_005902), SMAD4 (NM\_005359), *TAB2* (NM\_015093), *TGFB2* (NM\_001135599), *TGFB3* (NM\_003239), *TGFBR1* (NM\_001306210), *TGFBR2* (NM\_001024847), *UPF3B* (NM\_080632.2), and *ZDHHC9* (NM\_016032.3). Targeted fragments were then sequenced on a MiSeq platform (Illumina, San Diego, CA, USA) using a MiSeq Reagent kit V3 300 cycles flow cell. Data analysis was performed considering the frequency, impact on the encoded protein, conservation, and expression of variants using distinct tools, as appropriate ANNOVAR (http://annovar.openbioinformatics.org/en/latest/), dbSNP (https://www.ncbi.nlm.nih.gov/snp), 1000 Genomes (http://www.internationalgenome.org), ExAC (http://exac.broadinstitute.org). Selected variants are interpreted according to the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMGG/AMP) [12]. The identified variant was confirmed by Sanger sequencing. We checked the selected variant in the UMD-*FBN1* database, a computerized database that currently contains information about the published mutations of the *FBN1* gene.

#### *2.6. Variant Designation*

Nucleotide variant nomenclature follows the format indicated in the Human Genome Variation Society (HGVS, http://www.hgvs.org) recommendations. DNA variant numbering system refers to cDNA. Nucleotide numbering uses +1 as the A of the ATG translation initiation codon in the reference sequence, with the initiation codon as codon 1.

#### *2.7. In Silico Prediction*

In silico analysis of identified intronic variants was conducted by running three independent algorithms for splice signal detection: NetGene2 (http://www.cbs.dtu.dk/services/NetGene2/), Berkeley Drosophila Genome Project (BDGP, http://www.fruitfly.org/seq\_tools/splice.html), and Human Splicing Finder (HSF, http://www.umd.be/HSF3/).

#### *2.8. RNA Extraction and Reverse Transcription–Polymerase Chain Reaction*

Total RNA was extracted from lymphocytes fraction isolated from peripheral blood. Total RNA was extracted using RNeasy Mini Kit (Qiagen, Hilden, Germany), treated with DNase-RNase free (Qiagen, Hilden, Germany), quantified by Nanodrop (Thermo Fisher Scientific, Waltham, MA, USA), and reverse-transcribed using QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Primers used for DNA amplification and reverse transcription–polymerase chain reaction are reported in Table 1.

**Table 1.** Primers used for DNA amplification reverse transcription of *FBN1* fragments carrying the investigated variants (RefSeq NM\_000138) in individual 1 and individual 2.


#### **3. Results**

#### *3.1. Individual 1*

Molecular testing was carried out on the Individual 1 DNA by Sanger sequencing of the *FBN1* gene, and this gave normal results. Multiple ligation-dependent probe amplification analysis for *FBN1* intragenic deletions/insertions resulted normal. Given the convincing clinical picture and positive family history, re-analyzing raw data from the Sanger sequence of extended intronic sequences was carried out and revealed a novel *FBN1* heterozygous splicing variant c.6872-24T>A, p.(?) (Figure 2A). Co-segregation with the disease was demonstrated on the father's DNA.

Computational predictions conducted using NetGene2, BDGP, and HSF revealed that the intronic c.6872-24T>A variant might influence the splicing process by differentially affecting canonical versus cryptic splice site utilization. The newly detected variant was predicted to insert 22 nucleotides (nt) of intron 56 into the mature mRNA (Figure 2B). This should give rise to frameshift and a PTC (p.(Asp2291Glyfs\*9)), leading to the loss of the TGFβ binding protein homologous motifs, the last seven cysteine-rich EGF-like domains, and the FibuCTDIII-like motif.

**Figure 2.** Molecular analyses of individual 1. (**A**) Sanger sequencing in an unaffected (unrelated) control (CTRL), the proband and his father of the *FBN1* region spanning the 56 to 57 exon–intron junction. The *FBN1* genomic variant is indicated by an arrow. (**B**) *S*chematic representation of *FBN1* region including 55 to 57 exons and introns. Rectangles represent exons; thin horizontal lines represent introns. The position of the identified variant is indicated with an asterisk. Wild type and abnormal transcripts generated by the c.6872-24T>A splicing variant are shown. (**C**) Total RNA of a normal (unrelated) control, the proband, and his father were used for RT-PCR of the *FBN1* transcript. The same amounts of patient's and control cDNA were PCR amplified. PCR products were analyzed by electrophoresis on 3.5% agarose gel. M, DNA marker; C, control; P, proband; F, father. Affected individuals (lane P and F) present both the wild-type mRNA (lower band) and aberrantly spliced transcript (upper band). (**D**) Sanger sequencing of the *FBN1* transcript, including exons 56 to 57 in an unaffected (unrelated) control (C), the proband (P), and his father (F). Variant position on mRNA level was annotated (r.6871\_6872ins6872-22\_6872-1). The intronic region was reported at the bottom of the Sanger chromatograms of all affected individuals.

The effect of the *FBN1* splice site variant c.6872-24T>A at the mRNA level was ascertained by in vitro RT-PCR amplification of *FBN1* RNA of Individual 1 and his father. The sequencing of the RT-PCR products spanning the 55 to 57 exons of *FBN1* showed the insertion of 22 nts of intron 56–57 in the *FBN1* mRNA of Individual 1 and his father (r.6871\_6872ins6872-22\_6872-1) (Figure 2C,D). Electropherogram showed a significant reduction of the peaks corresponding to the mutated allele. The aberrantly spliced transcript is not present in the expressed sequence tag database (EST, https://www.ncbi.nlm.nih.gov/dbEST/index.html) of *FBN1*. Following available standards for the interpretation of sequence variants [12], the *FBN1* c.6872-24T>A variant was classified as likely pathogenic and submitted to the Leiden Open Variation Database (LOVD, https://www.lovd.nl; accession number: # 00229824).

#### *3.2. Individual 2*

Targeted NGS analysis revealed the *FBN1* heterozygous c.7571-12T>A variant in intron 61 (Figure 3A). No other candidate variant was found in the remaining genes. The c.7571-12T>A variant was not reported in the UMD-*FBN1* and was absent in the proband's healthy sister. Both parents died, and the other brothers and sisters were not available for genetic testing.

**Figure 3.** Molecular analyses of individual 2. (**A**) Sanger sequencing in an unaffected (unrelated) control and the proband of the *FBN1* region spanning 61 to 62 exon–intron junction. The *FBN1* genomic variant is indicated by an arrow. (**B**) *S*chematic representation of *FBN1* region including 61 to 62 exons and introns. Rectangles represent exons; thin horizontal lines represent introns. The position of the identified variant is indicated with an asterisk. Abnormal transcripts generated by the c.7571-12T>A splicing variant. (**C**) Total RNA of a normal (unrelated) control and the proband were used for RT-PCR of the *FBN1* transcript. The same amounts of patient and control cDNA were PCR amplified. PCR products were analyzed by electrophoresis on 3.5% agarose gel. M, DNA marker; C, control; P, the proband. M, DNA marker; C, control; P, proband; F, father. Affected individual (lane P) presents both wild-type (lower band) and aberrant transcript (upper band). (**D**) Sanger sequencing of the *FBN1* transcript, including exons 61 to 62 in an unaffected (unrelated) control (C) and the proband (P). Variant position on mRNA level was annotated (r.7570\_7571ins7571-10\_7571-1). The intronic region was reported at the bottom of the Sanger chromatogram of the affected individual.

The splicing predictor tools showed that the variant might alter the splicing process. In particular, all predicted with high probability that the c.7571-12T>A intronic variant activates a cryptic acceptor splice site within intron 61 that, in turn, leads to an inefficient recognition of the canonical splice site (Figure 3B) and insertion of intronic nucleotides in the *FBN1* mRNA. Sequence analysis on cDNA confirmed the c.7571-12T>A variant causes a shift from canonical toward cryptic splicing that leads to the inclusion of 10 nts of the 61 to 62 intronic sequence (r.7570\_7571ins7571-10\_7571-1) (Figure 3C,D) and a frameshift which generate a PTC in exon 61 (p.(Asp2524Argfs\*5)). The truncated protein loses the last seven cysteine-rich EGF-like domains and the FibuCTDIII-like motif. Electropherogram showed a reduction of the peaks corresponding to the mutated allele. The sequence of the alternative mRNA was absent in the EST database. The *FBN1* c.7571-12T>A variant was therefore interpreted as likely pathogenic and submitted to LOVD (accession number: #00229832).

#### **4. Discussion**

In this work, we added two novel intronic variants to the *FBN1* mutational repertoire and demonstrated their effect at the mRNA level. Our findings highlight the impact of variants affecting non-coding DNA in MFS and related disorders, and the opportunity to re-consider the standard diagnostic approaches, at least, in specific cases. In particular, molecular studies revealed that the novel c.6872-24T>A and c.7571-12T>A *FBN1* variants induce aberrant pre-mRNA splicing by the generation of a new cryptic acceptor site that outcompetes the canonical splice site and generate exonization of intronic sequences. This, in turn, results in a frameshift with the introduction of a PTC. In both cases, PTC is predicted to produce a truncated fibrillin 1 with a loss of critical C-terminal domains of fibrillin 1 [13].

The mechanism(s) by which *FBN1* splicing variants might exert their pathogenic effect is (are) complex, and both dominant negative and loss-of-function effects have been proposed based on the predicted consequences on mRNA, and protein structure and function(s) [14–16]. In splicing variants generating out-of-frame deletions/insertions, the introduction of a PTC and the synthesis of a truncated protein are assumed. However, PTC might elicit nonsense-mediated mRNA decay (NMD), with a variable proportion of the mutated allele acting as a *null* allele. If NMD affects ~100% of the transcripts, haploinsufficiency is the leading molecular pathogenesis. Conversely, the production of a truncated fibrillin 1 due to complete NMD escape, can impact its function due to truncated protein production. In proteins with pleiotropic effects and distinct domains with different functions, such as fibrillin 1 [3], the length of the synthesized truncated protein and the differential expression of the gene in the different tissues are the leading modulator factors of its residual functions. A mixed pathogenesis should also be considered for PTC that allow only in part NMD. In splicing variants leading to in-frame deletions/insertions, the predicted consequence might mirror the dominant negative effect attributed to point variants [17].

Literature on *FBN1* intronic variants affecting non-canonical splice sites is currently limited to single reports, and the emerging genotype–phenotype correlations are purely speculative. For example, Wang et al., by reporting a small case series of familial aortic aneurysms and dissection, studied the effect of the *FBN1* c.5917+6T>C variant in HEK293 cells and concluded that skipping of exon 47 seems to associate with a higher rate of cardiovascular involvement [18]. Wypasek and coworkers reported an adult with MFS, severe cardiac involvement but normally placed lens, and the *FBN1* c.1589-9T>A affecting splicing by in silico and in vitro investigations [19]. Recently, a further MFS patient with full cardiac, ocular, and skeletal phenotype has been reported with the *FBN1* c.2678-15C>A variant affecting splicing by minigene approaches [20].

In our study, both variants are predicted to result in a truncated protein. Although we did not further investigate their effect, both transcripts are likely destined, at least in part, to NMD, as the PTC falls in more than 50 nts upstream of the last exon–exon junction within the mRNA [21]. This might be hypothesized by the differences in peak heights in the electropherograms of *FBN1* transcripts (Figures 2 and 3). This result suggests that a portion of aberrantly *FBN1*-spliced transcripts could be degraded by NMD process in both cases. *FBN1* haploinsufficiency due to NMD-mediated degradation of aberrant mRNA might be the leading candidate molecular mechanism causing disease for the novel c.6872-24T>A and c.7571-12T>A variants. However, we cannot exclude that a proportion of the mutated transcripts escapes NMD at least in blood cells. Therefore, both NMD and the synthesis of a truncated protein may result from the identified causative variants. In Individual 1 the predicted truncated protein (p.(Asp2291Glyfs\*9)) losses the seven TGFβ binding protein homologous motifs, the last seven cysteine-rich EGF-like domains, and the FibuCTDIII-like motif; while in Individual 2, protein truncation should occur later (p.(Asp2524Argfs\*5)) preserving the loss of TGFβ binding protein homologous motifs. Individual 1 (and his affected father) showed severe aortic disease, a feature that was not observed in Individual 2, and both individuals lack ectopia lentis.

The reasons underlying the different clinical diagnoses in these two patients are hard to define. We can speculate that in Individual 1 aortic involvement might be, at least in part, due to the loss of TGFβ binding domain, which, in turn, is preserved in the truncated protein predicted in Individual 2 (MASS syndrome), who showed normal aorta. In fact, perturbed binding with TGFβ results in the alteration of elasticity and microfibrils stability, two factors primary influencing aortic wall integrity [22]. A trend toward an increased risk for ectopia lentis was reported in cysteine-destroying or cysteine-creating variants [2,15], a mechanism not easily evoked in our two Individuals. Therefore, our observation testifies once more the difficulties in tracing consistent genotype–phenotype correlations in MFS and the need for international collaborations working towards this aim.

In this work, we highlighted the importance of including extended intronic regions in the analysis of *FBN1* for clinical purposes. This remains valid in the era of NGS and should be considered at the stage of (virtual) multigene panel design, as well as during variant prioritization and post-analytical phase. Such an approach also needs the availability of additional tests validated for diagnostics and able to explore variant effects at the mRNA or post-transcriptional levels for supporting the clinical interpretation of the genomic data. Therefore, this work points out the need for introducing functional investigations in the diagnostic workflow of at least selected disorders or gene panels, for which a high rate of intronic variants with a potential pathogenic effect is expected. The description of further individuals with splicing variants with documented effects at the various post-genomic levels seems critical for improving our knowledge on this issue.

**Author Contributions:** Formal analysis, C.F.; Funding acquisition, L.M.; Investigation, C.F. and L.M.; Resources, S.M., A.F. and P.G.; Supervision, M.C.; Writing—original draft, C.F., P.G. and M.C.; Writing—review and editing, C.F., S.M., L.M., A.F., P.G. and M.C.

**Funding:** This work was supported by the Ricerca Corrente 2018–2021 Program from the Italian Ministry of Health.

**Acknowledgments:** The authors thank the families for their kind availability in sharing the findings within the scientific community The funders had no role in study design, data collection, and analysis, decision to publish, or preparation of the manuscript.

**Conflicts of Interest:** All authors declare no conflict of interest concerning this work.

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Genotypic Categorization of Loeys-Dietz Syndrome Based on 24 Novel Families and Literature Data**

**Letizia Camerota 1,**†**, Marco Ritelli 2,**†**, Anita Wischmeijer 3, Silvia Majore 4,5, Valeria Cinquina 2, Paola Fortugno 6, Rosanna Monetta 1,6, Laura Gigante 7,8,**

**Marfan Syndrome Study Group Tor Vergata University Hospital** ‡**, Federica Carla Sangiuolo 7,8, Giuseppe Novelli 7,8,9, Marina Colombi 2,\* and Francesco Brancati 1,6,\***


Received: 27 August 2019; Accepted: 21 September 2019; Published: 28 September 2019

**Abstract:** Loeys-Dietz syndrome (LDS) is a connective tissue disorder first described in 2005 featuring aortic/arterial aneurysms, dissections, and tortuosity associated with craniofacial, osteoarticular, musculoskeletal, and cutaneous manifestations. Heterozygous mutations in 6 genes (*TGFBR1*/*2, TGFB2*/*3, SMAD2*/*3*), encoding components of the TGF-β pathway, cause LDS. Such genetic heterogeneity mirrors broad phenotypic variability with significant differences, especially in terms of the age of onset, penetrance, and severity of life-threatening vascular manifestations and multiorgan involvement, indicating the need to obtain genotype-to-phenotype correlations for personalized management and counseling. Herein, we report on a cohort of 34 LDS patients from 24 families all receiving a molecular diagnosis. Fifteen variants were novel, affecting the *TGFBR1* (6), *TGFBR2* (6), *SMAD3* (2), and *TGFB2* (1) genes. Clinical features were scored for each distinct gene and matched with literature data to strengthen genotype-phenotype correlations such as more severe vascular manifestations in *TGFBR1*/*2*-related LDS. Additional features included spontaneous pneumothorax in *SMAD3*-related LDS and cervical spine instability in *TGFB2*-related LDS. Our study broadens the clinical and molecular spectrum of LDS and indicates that a phenotypic continuum emerges

as more patients are described, although genotype-phenotype correlations may still contribute to clinical management.

**Keywords:** hereditary connective tissue disorders; Loeys-Dietz syndrome; Ehlers-Danlos syndrome; arterial aneurysms; *TGFBR1*; *TGFBR2*; *SMAD2*; *SMAD3*; *TGFB2*; *TGFB3*

#### **1. Introduction**

Loeys–Dietz syndrome (LDS) is a rare hereditary connective tissue disorder (HCTD) with an autosomal dominant inheritance characterized by a widespread systemic involvement. The disorder was first described in 2005 and in its most typical presentation it features aortic/arterial aneurysms and/or dissections, as well as arterial tortuosity in association with variable craniofacial, osteoarticular, musculoskeletal, and cutaneous manifestations [1,2]. LDS is caused by pathogenic variants in transforming growth factor β (TGF-β) signaling pathway-related genes, i.e., *TGFBR1*, *TGFBR2*, *SMAD2*, *SMAD3*, *TGFB2,* and *TGFB3*, which alter the physiological development and function of the extracellular matrix, leading to cardiovascular and multisystem abnormalities [3]. Mutations in the genes encoding the TGF-β receptor subunits (*TGFBR1* and *TGFBR2*) were firstly identified and an initial clinical classification of LDS type I and type II was proposed based on the presence of typical craniofacial features in LDS type I [2]. Over the years, the discovery of novel disease-causative genes, eased by the advent of next-generation sequencing techniques, paved the way for novel genotype-phenotype correlations. For example, the identification of a subset of LDS patients with pathogenic variants in *SMAD3* displaying high frequency of osteoarthritis, prompted some authors to define the resulting phenotype as "aneurysms-osteoarthritis syndrome" or LDS type III [4]. McCarrick and coworkers argued that this phenotype fits within the LDS spectrum based on the analysis of large cohorts of mutated patients [5]. In particular, the authors concluded that, in the absence of formal diagnostic criteria for LDS, a pathogenic variant in any of the six LDS disease-causative genes in combination with the presence of arterial aneurysm or dissection or a positive family history should be considered sufficient for the diagnosis of LDS [5]. This stimulated the study of genotype-phenotype correlations, which are of paramount value for correct management of patients and to plan appropriate prevention programs in clinical practice. Even if wide interand intrafamilial variability in the distribution and severity of clinical features is observed, a more accurate gene-based categorization of LDS may have a great clinical impact in early diagnosis and management and guide treatment strategies for patients and family members with this life-threatening condition. Of note, the diagnosis of patients with LDS prompting molecular testing is not always straightforward, since the spectrum of clinical manifestations is broad and often overlaps other HCTDs, including Marfan syndrome (MFS), Shprintzen-Goldberg syndrome, some types of cutis laxa, Ehlers-Danlos syndromes (EDS) (particularly the vascular type), arterial tortuosity syndrome, congenital contractual arachnodactyly, and biglycan (*BGN*)-associated aortic aneurysm syndrome. Additionally, in some cases further genetic testing may be indicated for the differential diagnosis with other hereditary thoracic aortic disorders caused by pathogenic variants in distinct genes including *NOTCH1*, *ACTA2*, *MYH11*, *MYLK*, *PRKG1*, *MAT2A*, *FOXE3*, *MFAP5*, and *LOX* [6]. Herein, we describe 34 patients from 24 families with LDS in which we identified the genetic defect in one of the known genes; 15 variants were novel, expanding the mutational repertoire. We further strengthened genotype-phenotype correlations by fitting our data with available studies in literature and updated the gene-to-phenotype categorization of LDS.

#### **2. Patients and Methods**

#### *2.1. Patients*

Twenty-four index patients and 10 relatives with LDS, mostly of Italian origin but two from Philippines and Sri Lanka, were evaluated from 2010 to 2018 in specialized outpatient clinics for the diagnosis of HCTDs, namely: (*i*) the Ehlers-Danlos Syndrome and Inherited Connective Tissue Disorders Clinic at the University Hospital Spedali Civili of Brescia, (*ii*) the Medical Genetics Unit of the Sant'Orsola-Malpighi Hospital of Bologna, (*iii*) the Clinical Genetics Unit of the Regional Hospital of Bolzano, (*iv*) the Centre of Expertise for Marfan syndrome and Marfan-related disorders at Policlinico Tor Vergata University Hospital of Rome and (*v*) the Medical Genetics Unit, Department of Life, Health, and Environmental Sciences of the University of L'Aquila.

#### *2.2. Molecular Investigations*

Molecular analyses were performed in the laboratory of genetic testing at the Division of Biology and Genetics, Department of Molecular and Translational Medicine, University of Brescia. Mutational screening was achieved on genomic DNA purified from peripheral blood leukocytes of affected and unaffected family members by standard procedures. All exons and their intron-flanking regions of *TGFBR1* (NM\_004612.4, NP\_004603.1), *TGFBR2* (NM\_003242.5, NP\_0033233.4), *SMAD*3 (NM\_003242.5, NP\_005893.1), and *TGFB2* (NM\_001135599.3, NP\_001129071.1) were PCR amplified by using optimized genomic primers (Supplementary Table S1, primers were designed by using the Primer Express software v. 3.0.1 (Thermo Fisher Scientific, South San Francisco, CA, USA) and purchased by Metabion International AG, Planneg, Germany), which were analyzed for the absence of known variants using the GnomAD database [7]. After enzymatic cleanup of the PCR products, all fragments were sequenced in both orientations using the Big Dye Terminator Cycle Sequencing kit protocol (Thermo Fisher Scientific, South San Francisco, CA, USA) followed by capillary electrophoresis on the ABI3130XL Genetic analyzer. The sequences were analyzed with the Sequencher 5.0 software (Gene Codes Corporation, Ann Arbor, MI, USA) and variants were annotated according to the Human Genome Variation Society (HGVS) nomenclature with the Alamut Visual software version 2.11 (Interactive Biosoftware, Rouen, France), which was also used for splice site prediction, since it includes four different prediction algorithms: SpliceSiteFinder-like, MaxEntScan, NNSplice, and GeneSplicer. When available, fresh blood samples were collected in patients carrying variants affecting canonical splice sites. In order to verify the effect on splicing, RT-PCR was carried out on total RNA extracted from patients' whole blood stabilized in Paxgene tubes following the manufacturer's protocol by using the Paxgene Blood RNA Extraction Kit (PreAnalytiX, Qiagen, Hilden, Germany). In particular, amplification of cDNA covering exons 2–4 of *TGFBR2* (TGFBRex2-forw: 5 -GTGGCTGTATGGTAAGAGA-3 and TGFBR2ex4-rev: 5 -CCAGGTTGAACTCAGCTTCTG-3 ) and exons 6-8 of *SMAD3* (SMAD3ex6-forw: 5 -CCTAGGGCTGCTCTCCAATG-3 and SMAD3ex8-rev: 5 -GTGCACATTCGGGTCAACTG-3 ) was performed, respectively, and followed by Sanger sequencing of the RT-PCR products. All identified variants were submitted to the Leiden Open Variation Database (LOVD) [8].

#### *2.3. Genotype-Phenotype Analysis and Literature Review*

The medical records of 34 patients with a molecularly proven diagnosis of LDS were accessed and their phenotypes were defined using the Human Phenotype Ontology (HPO) terms by a careful review of clinical notes. Phenotypic categories were created based on the number of observations in the patient's cohort and their relationship to LDS. In reporting the frequency of clinical features of our case series, the used denominator refers to the number of subjects for which information on a specific feature was available (Table 1). We reviewed the medical literature on the clinical manifestations of patients with a confirmed genetic diagnosis of LDS, i.e., patients from large cohorts harboring mutations in one of the six known disease-causative genes. The primary search was completed in the PubMed database and limited to articles in the English language literature without restrictions on the date of publication. Keywords used were "Loeys-Dietz syndrome" AND "genotype", "Loeys-Dietz syndrome" AND "phenotype" as well as "Loeys-Dietz syndrome" AND "review." A secondary search was performed to identify pertinent articles cited in those selected in the primary search. We tried to identify all papers for which genotype-phenotype correlation data were available, focusing on reviews and large cohort studies.

#### *2.4. Ethical Compliance*

This study was approved by the relevant Ethical Authorities of the: Policlinico Tor Vergata University Hospital (Progetto di Ricerca PGR00229, R.S. 204/16); University of L'Aquila, Medical Genetics Section, Department of Life, Health, and Environmental Sciences (PGR00229, prot. 20251); and local Ethical Committees. This study was performed in agreement with the principles of the 1964 Helsinki Declaration. All subjects (or their legally authorized representative) enrolled into the genetic study provided written informed consent.

**Table 1.** Comparison of the clinical features observed in previously reported and LDS patients described in this cohort for each distinct


gene.

#### **3. Results**

#### *3.1. Demographic Data and Genotype-Phenotype Analysis of LDS Patient's Cohort*

We report the clinical and genetic features of 34 individuals from 24 families. The female-to-male ratio of affected probands was 1/1. The mean age at diagnosis of the 24 unrelated probands was 26 years (range 1–51 years). Thirteen of them had a positive family history, while 11 were sporadic. Ten individuals were diagnosed upon familial segregation. The detailed clinical and molecular features of each affected individual are summarized in Supplementary Table S2. The overall frequency of selected clinical features observed in our cohort, categorized for each distinct gene as compared to in the literature, is shown in Table 1. An overview of characteristic features observed in the patients of our cohort is presented in Figure 1.

**Figure 1.** Clinical and instrumental findings observed in our Loeys-Dietz syndrome (LDS) patients.

(**a**–**c**) Proband of family 1 (p.Asp400Gly,*TGFBR1*), aged 7 years manifesting dolichocephaly (**a**), milia (**b**) and enamel defect of permanent dentition. (**d**,**e**) Family 4 (p.Gly271Asp, *TGFBR1*), proband (**d**) and his brother (**e**) outlining intrafamilial variability of palpebral fissures, horizontal in the proband (**d**) and downslanted in his brother (**e**) in the presence of severe milia in both. (**f**) CT-angiography of the abdomen in the proband of family 7 (p.Gly353Arg, *TGFBR1*), aged 23 years, showing aneurysm of the abdominal aorta (asterisk) with intimal blister flap (arrow head). (**g**) Proband of family 15 (p.Asp522Asn, *TGFBR2*), aged 36 years: CT-angiography of head and neck showing fusiform aneurysm of the left subclavian artery (asterisk), middle cerebral artery stenosis (arrow heads), diffuse tortuosity of vertebral arteries (area within the circle). (**h**–**k**) Proband of family 16 (p.Asp446Asn, *TGFBR2*), aged 7 years, displays asymmetry of iliac crest, thin and translucent skin with highly visible subcutaneous venous reticulum (**h**); Limited function of the feet after corrective surgery for bilateral clubfoot (**i**); hind foot deformity with bilateral pronated valgus pes planus (**j**); X-ray frontal view performed at 6 years shows the displaced heart in the left thorax due to severe pectus excavatum and scoliosis with lumbar left rotation and asymmetric iliac wings (**k**). (**l**–**n**), Proband of family 24 (p.Phe160Leufs\*14, *TGFB2*), aged 48 years featuring facial asymmetry and dysmorphism (highly arched eyebrows, hypertelorism, bilateral exophthalmos, bifid nasal tip, long philtrum with thin upper lip, micrognathia) in addition to premature aging appearance (**l**); Joint laxity of the thumb (**m**) and the knee (**n**); Note the thin and translucent skin with visible subcutaneous veins.

#### *3.2. Molecular Findings*

In this study, 24 different variants in four distinct LDS-related genes were identified in the proband of each family by direct sequencing (Table 2). All variants were deposited in LOVD. The majority of variants affected the *TGFBR1* (9/24) and *TGFBR2* (10/24) genes, four families showed variants in *SMAD3*, and 1 patient harbored a *TGFB2* variant. No variants were identified in *SMAD2* and *TGFB3*. Fifteen variants affecting *TGFBR1*, *TGFBR2*, *SMAD3*, and *TGFB2* were novel, while nine in *TGFBR1*, *TGFBR2*, *SMAD3* were previously reported either in the literature or in the ClinVar database (Supplementary Table S2 and Table 2).

Among novel variants, 10 missense substitutions and one in-frame deletion of two highly conserved amino acid residues p.(Lys232\_Ile233del) were evaluated for their putative pathogenicity through different *in silico* prediction algorithms, all predicted them as being high impacting variants. Given that they are: (i) located in critical and well-established functional domains without benign variation; (ii) absent in publicly available population databases, (iii) predicted as deleterious my multiple lines of computational evidence and based on the matching of the clinical phenotypes, all these variants were classified as pathogenic (class 5) according to the American College of Medical Genetics and Genomics (ACMG) guidelines [9].

Likewise, the novel c.480del variant in *TGFB2* was classified as pathogenic (class 5), since it leads to frameshift and formation of a premature termination codon p.(Phe160Leufs\*14) that likely activates nonsense-mediated mRNA decay (NMD).

For two out of three variants affecting splice sites, we obtained fresh blood samples and demonstrated an effect on the splicing process, corroborating their pathogenicity. In particular, the c.263+6C>T variant in *TGFBR2* (rs758501054, Minor allele frequency in ExAC, MAF = 0.000008) was classified as likely pathogenic (class 4) according to the ACMG, since it creates a new splice donor site 6 bases downstream of the wild-type donor with retention of 4 nucleotides of intron 3, formation of a stop codon (p.Arg114\*) and activation of NMD. The c.1009+1G>A variant in *SMAD3* was classified as pathogenic (class 5), since it abolishes the canonical splice donor site of exon 7 causing in-frame exon skipping (p.Arg292\_Gly337del). Lastly, although RNA was not available to investigate its effect, the c.862\_871+8del pathogenic variant (class 5) variant in *SMAD3,* thought formally a frameshift variant, likely leads to abnormal splicing as well (i.e., in-frame exon 6 skipping), as the canonical splice donor site is lost.



#### **4. Discussion**

Since the relatively recent description of LDS in 2005, six disease-causative genes have been identified, namely *TGFBR1* (MIM #190181) [1], *TGFBR2* (MIM#190182) [1], *SMAD2* (MIM #601366) [10], *SMAD3* (MIM #603109) [4], *TGFB2* (MIM #190220) [11], and *TGFB3* (MIM#190230) [12]. This advised the subdivision of LDS into multiple classes based on the causative gene (LDS1-5) providing a general indication of the spectrum of disease severity, from most to least severe form: LDS1=LDS2>LDS3>LDS4>LDS5 [6]. Still, there are not enough data on LDS caused by heterozygous mutations in *SMAD2* to place this form (LDS6) in this spectrum [13]. Recently, it has been suggested that mutations in genes different from *TGFBR1*/2 may give rise to a phenotypic continuum hard to categorize in clear-cut genotype-phenotype correlations [3,6]. In this study, we performed a retrospective multi-center study on clinical and mutational analyses in 24 novel LDS families with 34 patients. Furthermore, a systematic overview of LDS patient's features for each mutated gene was conducted to verify if observed genotype-phenotype correlations fitted current knowledge.

Overall, we noticed significant overlaps for *TGFBR1*/*2*, *SMAD3,* and *TGFB2* patients (Table 1), while emerging genotype-phenotype correlations are described below, subdivided for each gene.

#### *4.1. TGFBR1*/*2 Genes*

According to previous studies, most of the identified mutations affected *TGFBR1* (9/24) and *TGFBR2* (10/24) genes. While nearly all mutated patients displayed ascending/aortic root aneurysm, severe arterial involvement was recorded in half. Early aortic dissection was more rarely reported together with arterial tortuosity. Interestingly, we noticed a high number of different arteries being affected. These findings highlight the need for extended arterial imaging, as was already recommended [5]. Valve abnormalities, in particular mitral valve prolapse and insufficiency, affected nearly half of the patients; isolated mitral valve prolapse was also present in seven out of 10 *TGFBR2* patients. As previously described, craniofacial features may be considered to be characteristics of *TGFBR1*/*2*-LDS including dolichocephaly, hypertelorism, malar hypoplasia and highly arched palate, while abnormal palate/bifid uvula was not as frequently found as described in the literature. Recurrent skeletal anomalies included scoliosis, pes planus, long slender fingers, marfanoid habitus, and pectus deformity. Hernias were of different types (diaphragmatic, inguinal, or umbilical) and together with joint laxity were registered in half of patients. Cutaneous findings appeared extremely frequent including translucent skin, easy bruising and striae; together with facial milia, described in half mutated patients, skin features appeared as extremely useful handle for diagnosis [14]. Overall, our data confirm broad overlap between LDS phenotypes caused by *TGFBR1*/*2* pathogenic variants, as previously assessed [6]. Also, in four families the presence of significant craniofacial malformations correlated with a higher risk of vascular manifestations [5,6]. Finally, two children had tooth abnormalities, in particular enamel defects, a possibly underestimated features in LDS in line with the role of TGF-β signaling in dental and enamel formation [15].

#### *4.2. SMAD3 Gene*

Among our three *SMAD3*-mutated families, early osteoarthritis, considered a hallmark of this subset of patients, was described only in one, although it segregated in the proband and his father. In line with our data, despite early studies reported osteoarthritis in all *SMAD3* patients [4], others noticed a lower incidence [16]. Interestingly, this family presented very mild cardiovascular anomalies, with only mitral valve prolapse registered in the proband. Generally, *SMAD3*-LDS patients displayed a less severe degree of cardiovascular complications. In fact, despite all the presence of thoracic aortic (and renal arteries) aneurysms, dissection was reported only once, as well as arterial tortuosity and varices. Additional clinical findings overlapped with those recorded in other LDS patients, including recurrent craniofacial dysmorphism such as hypertelorism, malar hypoplasia, micrognathia and highly arched palate, cleft palate and/or uvula, cutaneous features with translucent skin, striae and facial milia,

hernias and skeletal manifestation like pes planus, pectus deformity, marfanoid habitus, and scoliosis. It is worth noting that one subject had spontaneous pneumothorax, recorded in a single *SMAD3* patient in the literature [3]. The causes of spontaneous pneumothorax in LDS have not been clearly elucidated but its association with LDS independently from the genetic subtype as well as the occurrence in other HCTD such as Marfan syndrome argues in favor of a common role for TGF-beta signaling. Several lines of evidence based on mouse models, knockout for genes encoding members of the TGF-beta pathway, support this observation (reviewed in reference [17]). In particular, perturbed TGF-beta signaling was implicated in abnormal pulmonary alveolarization and alveolar destruction, leading with age to emphysema and spontaneous lung rupture (i.e., primary spontaneous pneumothorax).

#### *4.3. TGFB2 Gene*

We identified only one patient with a *TGFB2* pathogenic variant, whose cardiovascular features consisted in mitral valve prolapse and insufficiency, mild arterial tortuosity and varicose veins, in the absence of aneurysms. Additional features included bifid uvula, striae, translucent and hyperextensible skin, facial milia, hernias, joint hypermobility, pectus deformity, scoliosis, and pes planus observed in other LDS genetic subtypes. In line with previous observations, this patient presented the mildest end of LDS phenotypic spectrum [18]. However, it should be noted that phenotypic variability ranging from mild to severe expression was also recorded in *TGFB2*-LDS [19]. Further broadening the clinical picture of this genetic subtype, she also showed cervical spine instability, which was not previously registered in association with *TGFB2*, although it has been commonly encountered in LDS [3]. Interestingly, in Tgfb2 heterozygous knockout mice severe skeletal defects are also observed including skull base and vertebral malformations, in line with the human phenotype [20]. These data corroborate the importance to investigate and diagnose malformation of cervical spine and/or instability, which should be carefully monitored to prevent severe complications in these patients.

#### *4.4. Other Genes*

No mutations were identified in the latest described LDS genes, namely *SMAD2* and *TGFB3*. Still, limited numbers are available in literature about genotype-phenotype correlation. To date, 9 (likely) pathogenic variants in *SMAD2* have now been described in 15 subjects displaying a broad range of features, including aneurysms, tortuosity of the entire arterial tree, and coronary artery dissections, even in the absence of prominent connective tissue characteristics [13]. Concerning *TGFB3*, 15 different variants were reported in 56 individuals presenting with phenotypic overlap between LDS and MFS [12].

#### **5. Conclusions**

In this work, we broadened the clinical and molecular spectrum of LDS, corroborated, and expanded previously delineated genotype-phenotype correlations, paving the way for a gene-based classification of different disease subtypes. Larger cohort screenings are needed to unravel the thorough clinical and molecular repertoires in LDS to accurately establish diagnostic criteria, define genotype-phenotype correlations, and collect natural history data for clinical prognostication.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/10/764/s1, Table S1: Primers, Table S2: Clinical and molecular features of LDS patients described in this cohort.

**Author Contributions:** All authors fulfill the criteria for authorship. F.B. and M.C. conceived the study and wrote the manuscript drafted by L.C. and M.R. who organized data contents, reviewed the literature and performed the tables; F.B. and L.C. prepared the figure; F.B., L.C. and L.G. performed clinical evaluation and follow-up of families 1, 2, 3, 4, 5, 10, 20; F.C.S. and G.N. performed genetic counseling and reviewed molecular data for families 1, 2, 3, 4, 5, 10, 20; A.W. performed clinical evaluation, follow-up and genetic counseling for families 6, 8, 16, 17, 18, 19, 21, 22, 23; S.M. performed clinical evaluation and follow-up of families 7 and 15; M.C. and M.R. performed genetic counseling and follow-up for families 9,11, 12, 13, 14, 22, 24; M.C., M.R. and V.C. carried out molecular investigations and RNA studies; R.M. and P.F. reviewed, analyzed and interpreted molecular data; L.C. submitted the study. All authors discussed, read, and approved the final manuscript.

**Funding:** F.B. receives funding from the University of L'Aquila (Fondi RIA 2019 and Fondi Ricerca Premiale 2017), the Italian Ministry of Health (Ricerca Finalizzata GR-2013-02356227) and the Undiagnosed Disease Network Italy established at Istituto Superiore di Sanità, Italy (Farmindustria, contributo incondizionato FAC. X7E).

**Acknowledgments:** The authors wish to thank all patients for their cooperation during the diagnostic process. M.R, V.C., and M.C. thank the Fazzo Cusan family for its generous support. We thank Domenica Taruscio and Marco Salvatore from the Istituto Superiore di Sanità (Italy) for their constant support.

**Conflicts of Interest:** All authors declare that there is no conflict of interest concerning this work.

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Genes* Editorial Office E-mail: genes@mdpi.com www.mdpi.com/journal/genes

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18