**Molecular Genetics and Pathogenesis of Ehlers-Danlos Syndrome and Related Connective Tissue Disorders**

Special Issue Editors **Marina Colombi Marco Ritelli**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Special Issue Editors* Marina Colombi University of Brescia Italy

Marco Ritelli University of Brescia Italy

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Genes* (ISSN 2073-4425) (available at: https://www.mdpi.com/journal/genes/special issues/Connective Tissue Disorders).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Article Number*, Page Range.

**ISBN 978-3-03936-322-3 (Hbk) ISBN 978-3-03936-323-0 (PDF)**

c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**




### **About the Special Issue Editors**

**Marina Colombi**, PhD, is a clinical and molecular geneticist who specializes in heritable disorders of connective tissue. She graduated with honors in Biological Sciences, option Genetics, at the University of Pavia, Italy, and she completed a PhD in genetic engineering at the University of Helsinki, Finland. She was research assistant professor of Applied Biology at the School of Medicine, University of Brescia, Italy, where she became full professor of Medical Genetics. Prof Colombi is the Director of the Division of Biology and Genetics, Department of Molecular and Translational Medicine and of the Postgraduate School of Medical Genetics. She is the director of the Ehlers–Danlos Syndrome and Inherited Connective Tissue Disorders Outpatient Clinic (CESED) at the Spedali Civili University Hospital of Brescia and of the Observatory on Connective Tissue Disorders (OCE) at the University of Brescia. She has various responsibilities at the University of Brescia, with a focus on the clinical genetics of rare diseases and genetic laboratory testing. Her major diagnostic and research interests include the clinical and molecular characterization of patients affected with several connective tissue disorders, and the definition of associated pathogenetic mechanisms. Most recently, her group characterized large cohorts of patients affected with classical and vascular Ehlers-Danlos syndrome as well as with ultrarare Ehlers-Danlos syndromes, arterial tortuosity syndrome, Loeys-Dietz syndrome, Marfan syndrome, cutis laxa, pseudoxanthoma elasticum, osteogenesis imperfecta, and dystrophic epidermolysis bullosa. The work of her group contributed to the definition of disease mechanisms involved in the pathogenesis of hypermobile, vascular, and classical Ehlers-Danlos syndrome, and arterial tortuosity syndrome by studying transcriptome profiling, extracellular matrix alterations, and aberrant signal transduction pathways in skin fibroblasts from patients affected with these disorders. She authored/co-authored 160 articles in international journals and various book chapters and books.

**Marco Ritelli**, PhD, is a clinical molecular geneticist who specializes in heritable disorders of connective tissue. He graduated with honors in Biological Sciences, option Genetics, at the University of Padua, Italy, and, at the Division of Biology and Genetics, Department of Molecular and Translational Medicine, of the University of Brescia, he performed specific training in human and molecular genetics as a research fellow, mainly focusing on the molecular characterization of patients affected with Ehlers-Danlos syndrome and related heritable connective tissue disorders and on the study of the pathomechanisms of these rare diseases. After obtaining his residency in Medical Genetics and the qualification as research assistant professor at the School of Medicine, University of Brescia, Italy, he currently has full-time involvement in diagnostic and research activity at the same institution within the group of Prof. Marina Colombi. Overall, his activities resulted in 80 peer-reviewed publications, most of which are on clinical and molecular research on Ehlers-Danlos syndrome and related disorders.

### *Editorial* **Molecular Genetics and Pathogenesis of Ehlers–Danlos Syndrome and Related Connective Tissue Disorders**

#### **Marco Ritelli and Marina Colombi \***

Division of Biology and Genetics, Department of Molecular and Translational Medicine, University of Brescia, 25123 Brescia, Italy; marco.ritelli@unibs.it

**\*** Correspondence: marina.colombi@unibs.it; Tel.: +39-03-0371-7240

Received: 8 May 2020; Accepted: 11 May 2020; Published: 13 May 2020

**Abstract:** Ehlers–Danlos syndromes (EDS) are a group of heritable connective tissue disorders (HCTDs) characterized by a variable degree of skin hyperextensibility, joint hypermobility and tissue fragility. The current EDS classification distinguishes 13 subtypes and 19 different causal genes mainly involved in collagen and extracellular matrix synthesis and maintenance. EDS need to be differentiated from other HCTDs with a variable clinical overlap including Marfan syndrome and related disorders, some types of skeletal dysplasia and cutis laxa. Clinical recognition of EDS is not always straightforward and for a definite diagnosis, molecular testing can be of great assistance, especially in patients with an uncertain phenotype. Currently, the major challenging task in EDS is to unravel the molecular basis of the hypermobile EDS that is the most frequent form, and for which the diagnosis is only clinical in the absence of any definite laboratory test. This EDS subtype, as well as other EDS-reminiscent phenotypes, are currently investigated worldwide to unravel the primary genetic defect and related pathomechanisms. The research articles, case report, and reviews published in this Special Issue focus on different clinical, genetic and molecular aspects of several EDS subtypes and some related disorders, offering novel findings and future research and nosological perspectives.

**Keywords:** Ehlers–Danlos syndrome; heritable connective tissue disorders; differential diagnosis; next generation sequencing (NGS); transcriptomics; integrated omics approaches

Ehlers–Danlos syndromes (EDS), with an estimated prevalence of about 1/5000, belong to the large group of heritable connective tissue disorders (HCTDs) and are characterized by a variable degree of skin hyperextensibility, joint hypermobility (JHM), and tissue fragility. The clinical and genetic heterogeneity of these conditions has long been recognized, but the subjective interpretation of some semiquantitative clinical signs, such as skin hyperextensibility, skin texture, JHM, tissue fragility and bruising, led to diagnostic ambiguity and confusion regarding the type of EDS and the inclusion of similar phenotypes under the broad diagnosis of EDS. With more systematic research on clinical data and with the clarification of the molecular basis and associated pathomechanisms of several of these EDS phenotypes, different classification systems have been formulated in the past 50 years. A first classification with five main types was introduced in 1970 by Beighton [1], followed by the Berlin classification with 11 types [2], and the Villefranche nosology of 1997, in which six EDS types were included [3]. The rapid development of genetic techniques has allowed the recognition of many distinct disorders that, while dissimilar from the initially-described classic EDS types, have been given the umbrella term of EDS as an image of the presence of generalized connective tissue fragility. Hence, the last version of the EDS classification published in 2017, which recognized 13 types with 19 different causal genes mainly involved in collagen and extracellular matrix (ECM) synthesis and maintenance, has been expanded to include a wide range of clinically heterogenous disorders [4]. It should be noted that in an important percentage of EDS patients, no pathogenic variant in any of the known EDS-associated genes is identified. Therefore, it is expected that by taking advantage of next generation sequencing (NGS) technologies, further EDS types will be molecularly defined, thus demanding updating of the existing classification.

It always has been, and still is, a challenge to classify single patients in one of the existing EDS subtypes because the currently defined clinical criteria remain relatively unspecific. Often it is not possible to reach a clinical diagnosis and, therefore, the identification on molecular genetic testing of a clear pathogenic variant in a specific gene can be of great assistance, especially in patients with a clinical presentation that does not completely fit into one of the existing subtypes. The difficulty of clinical diagnosis is, among other reasons, due to the clinical overlap not only between many of the EDS subtypes but also with other HCTDs, such as Marfan, Loeys–Dietz, and arterial tortuosity syndromes, as well as some types of skeletal dysplasia and cutis laxa [4–10].

The major challenging task in EDS today is to unravel the molecular basis of the most frequent EDS type, namely hypermobile EDS (hEDS), the diagnosis of which remains reliant on clinical findings for the absence of any definite laboratory test. This EDS subtype as well as other EDS-reminiscent phenotypes are currently investigated worldwide to unravel the primary genetic defect and related pathomechanisms. In 2018, the groundbreaking "Hypermobile Ehlers Danlos Genetic Evaluation" (HEDGE) was launched by the International EDS Society (https://www.ehlers-danlos.com). There has never been such a worldwide collaborative effort before dedicated to discovering the underlying genetic markers for hEDS. Until the end of 2020, the HEDGE study aims to recruit, screen and undertake NGS on 1000 individuals who have been diagnosed with hEDS by the most recent clinical criteria established in 2017, which are stricter than the Villefranche criteria, in order to form homogeneous cohorts for research purposes [4]. Understanding the genetic causes of hypermobile EDS is undeniably central to the EDS community, since it will allow us to make unequivocal diagnoses for a huge number of patients. Furthermore, understanding the genetic pathways and etiopathomechanisms leading to hEDS will advise the search for possible therapeutic approaches for this disorder.

The original research articles, case report and reviews published in this Special Issue focus on different clinical, genetic, biological and molecular aspects of several EDS subtypes and some related disorders. When the first edition of McKusick's book entitled "Heritable disorders of connective tissue" was published in 1956, less than 100 manuscripts had been dedicated to EDS; they were mainly case reports [11]. Nowadays, the search term "Ehlers–Danlos syndrome" in PubMed yields more than 4000 papers including the 14 contributions of this Special Issue, demonstrating how significant and fertile scientific research is for rare genetic diseases such as EDS and related HCTDs.

The relevance of clinical and genetic research in the continuous definition of new EDS types is exemplified in the original research article by Ritelli et al. [12], who describe a novel patient with the classical-like EDS type 2 (clEDS 2) that is caused by recessive variants in *AEBP1*, and review the clinical and molecular findings of the few patients reported to date. This rare EDS type in differential diagnosis with the more frequent classical EDS (cEDS) is not yet included in the current EDS classification, since the first description was noted in 2018 [13], but certainly it will be incorporated in the forthcoming revision of the EDS nosology. Two original research articles, respectively by Rymen et al. [14] and Micale et al. [15], further define the phenotype of the other classical-like EDS type (*TNXB* deficiency) by reporting three novel patients and performing a literature review. The authors highlight that clEDS 1 is likely underdiagnosed due to the complex structure of the *TNXB* locus which complicates diagnostic molecular testing. Rymen et al. [14] also provide an in vitro characterization of the clEDS 1 cellular phenotype, demonstrating the disorganization of the type I, III and V collagen ECMs in patient's fibroblasts. The case report by Angwin et al. [16] underscores the importance of molecular analysis for a definite cEDS diagnosis by showing that patients with pathogenic *COL5A1* variants can have an absence of collagen flowers on skin biopsy transmission electron microscopy (TEM) analysis, which for many years has been recommended as a first line of investigation to confirm or exclude a cEDS diagnosis. The original research article of Miller et al. [17] provides novel clinical and instrumental

findings on hEDS, cEDS, and vascular EDS (vEDS) by performing the assessment of pulse wave velocity measurement in a large patient cohort, which is recognized as a gold standard for determining the stiffness of arteries. The authors evidenced an increased arterial elasticity in all EDS subtypes that was associated with lower supine and seated systolic and diastolic blood pressure, thus likely contributing to the orthostatic symptoms frequently encountered in EDS, especially hEDS.

The paper by Chiarelli et al. [18] offers a wide overview on molecular mechanisms likely involved in cEDS, vEDS, and hEDS that could direct future studies to possible therapeutic strategies. The authors review their previous transcriptome and protein studies on patient dermal fibroblasts, emphasizing that these cells, despite sharing a common ECM remodeling, show differences in the underlying pathomechanisms. In cEDS and vEDS fibroblasts, key processes such as collagen biosynthesis/processing, protein folding quality control, endoplasmic reticulum (ER) homeostasis, autophagy, and wound healing emerged as perturbed. In hEDS cells, gene expression changes related to cell–matrix interactions, inflammatory/pain responses, and the acquisition of an in vitro pro-inflammatory myofibroblast-like phenotype seem to contribute to the complex pathogenesis of this molecularly unsolved EDS type. The evidence that the application of untargeted general omics approaches may serve as a valuable tool to identify novel proteins or pathways involved in the pathogenesis of the different EDS types is also documented in the original research article by Lim et al. [19] that focuses on the very rare kyphoscoliotic EDS (kEDS) type, which groups two clinically indistinguishable disorders caused by biallelic variants in either *PLOD1* or *FKBP14.* This article also proves that nowadays it is possible to perform a high-profile scientific investigation for very rare genetic disorders which have been neglected for too long. The authors performed transcriptome profiling by RNA sequencing of kEDS patient-derived skin fibroblasts that revealed the differential expression of genes encoding ECM components that are unique between *PLOD1*-kEDS and *FKBP14*-kEDS, as well as genes involved in inner ear development, vascular remodeling, ER stress and protein trafficking that were differentially expressed in patient cells compared to controls, addressing possible pharmacological targets to improve disease symptoms.

We recommend reading two papers of the Special Issue together, namely first the research article by Caraffi et al. [20] and then the paper by Ritelli et al. [21], since they present stimulating results offering nosological viewpoints concerning the so-called linkerophaties (LKs), which are caused by defects in genes involved in the glycosaminoglycan (GAG) biosynthesis. Specifically, LK genes encode for enzymes that add GAG chains onto proteoglycans via a common tetrasaccharide linker region. LKs include two different subtypes of the spondylodysplastic EDS (spEDS type 1 and 2) and further related disorders that are characterized by a variable mixed phenotype with signs of EDS and skeletal dysplasia. Of note, some of these conditions are in fact included either in the 2017 EDS classification [4] or in the 2019 nosology of skeletal dysplasia [10]. In the original research article by Caraffi et al. [20], the clinical and molecular findings of three spEDS patients are reported. Through the description of one patient with *B4GALT7*- and two patients with *B3GALT6*-spEDS and a review of previous literature reports, the authors contribute to a more accurate definition of the clinical features associated with these rare conditions. Ritelli et al. [21] report on a patient fulfilling the diagnostic criteria for spEDS according to the 2017 nosology, in whom, however, NGS identified compound heterozygosity for two pathogenic variants in *B3GAT3* that is not recognized as an EDS-causing gene. The authors review the spectrum of *B3GAT3*-related disorders and provide a comparison of all LK patients reported at the time of writing, corroborating the notion that LKs are a phenotypic continuum bridging EDS and skeletal disorders. Following these papers, we suggest reading the research article by Kumps et al. [22] that further accentuates the existing nosological confusion concerning these very rare syndromes with huge clinical overlap. Indeed, the authors describe four patients with recessive variants in *SLC39A13* that are associated with spEDS type 3, even if the encoded gene product, namely a putative zinc transporter (contrariwise to the proteins encoded by *B4GALT7* and *B3GALT6*), is not involved in GAG biosynthesis. Given that the clinical presentation of this condition in childhood consists mainly of short stature and characteristic facial features, the authors propose that the differential diagnosis is not necessarily that of

a connective tissue disorder and that *SLC39A13* should be included in gene panels designed to address dysmorphism and short stature. In the outstanding review by Kosho et al. [23], the authors discuss recent advances in the pathophysiology of the rare musculocontractural EDS (mcEDS), which is caused by biallelic variants in *CHST14* and *DSE* (which are also involved in GAG synthesis). By describing novel glycobiological, pathological, and animal model-based findings, the authors highlight the critical roles of dermatan sulfate (DS) and DS-proteoglycans in the multisystem development and maintenance of connective tissues and provide fundamental evidence to support future etiology-based therapies.

Finally, three papers deal with different HCTDs in differential diagnosis with EDS. Beyens et al. [24] describe the clinical and molecular characteristics of two novel and 32 previously reported patients with occipital horn syndrome (OHS), previously known as EDS type IX or X-linked cutis laxa [2], caused by pathogenic variants in *ATP7A* (encoding a copper transporter). The main clinical features of OHS, such as cutis laxa, bony exostoses and bladder diverticula, are attributed to defective ATP7A trafficking and decreased activity of lysyl oxidase, a cupro-enzyme involved in collagen crosslinking, in line with a pathogenetic scheme shared with many EDS types. The authors explored the pathomechanisms of OHS by performing TEM analysis on skin biopsies and collagen biochemical analysis on fibroblast cultures that showed increased collagen diameter, elastic fiber abnormalities and multiple autophagolysosomes. Fusco et al. [25] report two unrelated individuals with Marfan syndrome (MFS) and Mitral valve–Aorta–Skeleton–Skin (MASS) syndrome, respectively, which were associated with different intronic variants in *FBN1*, pointing out the importance of intronic sequence analysis and the need for integrative functional studies in the diagnosis of MFS and related disorders. Camerota et al. [26] report on a cohort of 34 patients with Loeys–Dietz syndrome (LDS) with a defined molecular defect either in *TGFBR1*, *TGFBR2*, *SMAD3,* or *TGFB2*. The study broadens the clinical and molecular spectrum of LDS, corroborates and expands previously delineated genotype–phenotype correlations, and shows that a phenotypic continuum emerges as more patients are described, paving the way for a gene-based classification of the different disease subtypes.

In conclusion, this Special Issue, by offering novel findings and future research perspectives, will be of interest not only to a wide range of investigators but also to patients with EDS and related disorders, as well as to all healthcare practitioners who may encounter such syndromes during their work and we hope they will enjoy reading it.

**Author Contributions:** M.R. and M.C. equally contributed to the managing of the Special Issue and wrote this editorial. All authors have read and agreed to the published version of the manuscript.

**Funding:** No funding was active on this project.

**Acknowledgments:** M.R. and M.C. would like to thank the Editor-in-Chief and the Section Editors of Genes for the fruitful collaboration and all the authors and reviewers for their contributions to this Special Issue.

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

#### **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/).

*Article*

### **Expanding the Clinical and Mutational Spectrum of Recessive** *AEBP1***-Related Classical-Like Ehlers-Danlos Syndrome**

### **Marco Ritelli 1, Valeria Cinquina 1, Marina Venturini 2, Letizia Pezzaioli 1,3, Anna Maria Formenti 4, Nicola Chiarelli <sup>1</sup> and Marina Colombi 1,\***


Received: 10 January 2019; Accepted: 8 February 2019; Published: 12 February 2019

**Abstract:** Ehlers-Danlos syndrome (EDS) comprises clinically heterogeneous connective tissue disorders with diverse molecular etiologies. The 2017 International Classification for EDS recognized 13 distinct subtypes caused by pathogenic variants in 19 genes mainly encoding fibrillar collagens and collagen-modifying or processing proteins. Recently, a new EDS subtype, i.e., classical-like EDS type 2, was defined after the identification, in six patients with clinical findings reminiscent of EDS, of recessive alterations in *AEBP1,* which encodes the aortic carboxypeptidase–like protein associating with collagens in the extracellular matrix. Herein, we report on a 53-year-old patient, born from healthy second-cousins, who fitted the diagnostic criteria for classical EDS (cEDS) for the presence of hyperextensible skin with multiple atrophic scars, generalized joint hypermobility, and other minor criteria. Molecular analyses of cEDS genes did not identify any causal variant. Therefore, *AEBP1* sequencing was performed that revealed homozygosity for the rare c.1925T>C p.(Leu642Pro) variant classified as likely pathogenetic (class 4) according to the American College of Medical Genetics and Genomics (ACMG) guidelines. The comparison of the patient's features with those of the other patients reported up to now and the identification of the first missense variant likely associated with the condition offer future perspectives for EDS nosology and research in this field.

**Keywords:** classical Ehlers-Danlos syndrome; classical-like Ehlers-Danlos syndrome type 2; *AEBP1*; aortic carboxypeptidase-like protein; differential diagnosis; high-frequency ultrasonography; reflectance confocal microscopy

#### **1. Introduction**

Ehlers-Danlos syndrome (EDS), with an estimated prevalence of 1/5000, comprises a group of clinically heterogeneous heritable connective tissue disorders (HCTDs) with diverse molecular etiologies. The 2017 revised EDS classification recognized 13 distinct subtypes caused by pathogenic variants in 19 genes mainly encoding fibrillar collagens, collagen-modifying proteins, or processing enzymes [1]. Classical EDS (cEDS) (MIM #130000), with an estimated prevalence of 1/20,000, is an autosomal dominant disorder primarily characterized by cutaneous and articular involvement. Indeed, cEDS is suggested by skin hyperextensibility plus atrophic scarring that must be present together with the other major criterion, i.e., generalized joint hypermobility (gJHM) evaluated

according to the Beighton score (BS ≥5/9), and/or with at least three of the minor criteria among easy bruising, soft, doughy skin, skin fragility, molluscoid pseudotumors, subcutaneous spheroids, hernia (or a history of thereof), epicanthal folds, JHM complications (e.g., sprains, luxation/subluxation, pain, flexible flatfoot), and family history of a first-degree relative who meets clinical criteria [1–3]. Furthermore, cEDS patients may present distinctive facial features, premature rupture of fetal membranes, scoliosis, osteoporosis, gastroesophageal reflux, and cardiac and blood vessel fragility [2,4–8]. Skin is hyperextensible if it can be stretched over a standardized cut off in the following areas: 1.5 cm for the distal part of the forearms and the dorsum of the hands; 3 cm for neck, elbow and knees; 1 cm on the volar surface of the hand (palm) [1,2,5]. Atrophic scarring can range in severity; however, most cEDS patients have wide atrophic scars in different body areas that can variably assume a cigarette paper, papyraceous, or hemosiderotic appearance [1,2,5].

Point mutations or intragenic rearrangements of the *COL5A1* and *COL5A2* genes encoding type V collagen are recognized in over 90% of patients [4,9], the recurrent heterozygous *COL1A1* c.934C>T (p.Arg312Cys) substitution is rarely found [4,10,11]. Negative molecular testing does not exclude the diagnosis, as specific types of mutations (e.g., deep intronic variants) may go undetected by standard diagnostic molecular techniques. Nevertheless, alternative diagnoses should be taken into account in the absence of a *COL5A1, COL5A2,* and *COL1A1* mutation [1].

Recognition of cEDS is straightforward in the patient with the typical cutaneous signs and BS ≥5. However, intra- and interfamilial variability tells a much broader clinical presentation and significant overlap with other EDS types and HCTDs [1,2,4–6,12]. Differential diagnosis of cEDS should include the hypermobile EDS (hEDS), particularly in patients without a striking cutaneous involvement [1,12]. Indeed, hEDS shares with cEDS gJHM and many mucocutaneous signs, but generally a lower grade of skin hyperextensibility and only few small atrophic or post-surgical enlarged scars are observed [13,14]. In case of a family history compatible with autosomal recessive transmission, differential diagnosis comprises the rare classical-like EDS type 1 (MIM #606408) due to biallelic *TNXB* mutations. These patients show marked skin hyperextensibility, easy bruising, and joint laxity, but unlike cEDS patients, they do not have atrophic scarring or poor wound healing. Furthermore, minor criteria such as foot deformities, edema in the legs, mild proximal and distal muscle weakness, axonal polyneuropathy, and atrophy of muscles in hands and feet facilitates the differential [1,11]. Severe progressive cardiac-valvular problems distinguish the cardiac-valvular EDS type (*COL1A2*) from cEDS, severe skin fragility and unusual craniofacial features discriminates the dermatosparaxis EDS (*ADAMTS2*), whereas (congenital) kyphoscoliosis and muscle hypotonia differentiates the kyphoscoliotic EDS (*PLOD1*, *FKBP14*), which are other rare recessive EDS types. Bilateral congenital hip dislocation differentiates the autosomal dominant arthrochalasia EDS (*COL1A1*, *COL1A2*) [1,11,12].

Recently, in six individuals from four unrelated families who presented with a constellation of clinical findings reminiscent of cEDS such as gJHM, redundant and hyperextensible skin with poor wound healing and abnormal scarring [15–17], and recessive alterations in the *AEBP1* gene, which encodes the aortic carboxypeptidase-like protein (ACLP) associating with collagens in the extracellular matrix, were recognized, thus defining a new EDS form labelled as classical-like EDS type 2 (MIM #618000).

Herein, we describe an additional patient with a homozygous missense *AEBP1* causative variant and compare her clinical features with those of the other patients reported so far, offering future perspectives for EDS nosology and research in this field.

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

#### *2.1. Molecular Analyses*

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 the Italian legislation on genetic diagnostic tests and the patient provided written informed consent for publication of clinical data and photographs according to the Italian bioethics laws. Since this report is based on data obtained through routine clinical care and is not considered research at the involved institutions; formal ethics review was not obtained. Genomic DNA was extracted from peripheral blood leukocytes using standard procedures; the exons and intron-flanking regions of *COL5A1*, *COL5A2*, and exon 14 of *COL1A1* (c.934C>T (p.Arg312Cys) were amplified by PCR and directly sequenced using an ABI PRISM® 3130XL Genetic Analyzer (Life Technologies, Carlsbad, CA, USA), as previously reported [4]. For the multiplex ligation-dependent probe amplification (MLPA), the commercially available SALSA MLPA kits P331 and P332 for *COL5A1* gene were used, according to the manufacturer's recommendations (MRC-Holland, Amsterdam, The Netherlands), as previously described [4]. The primers for *AEBP1* Sanger sequencing (Supplementary Table 1) were designed for all coding exons, including the intron-exon boundaries, and primer sequences were analyzed for the absence of known variants using the GnomAD database [18]. The sequences were analyzed with the Sequencher 5.0 software and variants were annotated according to the Human Genome Variation Society (HGVS) nomenclature by using the Alamut Visual software version 2.11. To evaluate the putative pathogenicity of the *AEBP1* missense variant, which was submitted to the LOVD Ehlers–Danlos Syndrome Variant Database [19], we used the following mutation prediction programs: Mutation Assessor [20], PhD-SNP [21], Align GVD [22], SIFT [23], Mutation Taster [24], PolyPhen2 [25], PROVEAN [26], MutPred [27], M-CAP [28], CADD [29], DANN [30], Fathmm-MKL [31], and VEST [32]. The nucleotide and protein accession numbers correspond to the *AEBP1* (NM\_001129.4, NP\_001120.3) reference sequences.

#### *2.2. High-Frequency Ultrasonography and In Vivo Reflectance Confocal Microscopy*

To investigate patient's skin by a non-invasive approach, we performed high-frequency ultrasonography (HF-USG) and *in vivo* reflectance confocal microscopy (RCM) as previously described [33–35].

Briefly, HF-USG was performed on the dorsal and volar side of the forearm of the patient and 10 age- and gender-matched healthy individuals with the same skin phototype and similar sun exposure history by digital 50-MHz ultrasonography B mode scanning (DUB-USB Skin Scanner, Taberna Pro Medicum Company, Lueneburg, Germany). For ultrasound transmission, water was employed as a coupling medium between the transducer and the skin surface. The usable depth of signal penetration was 4 mm, and the gain was 40 dB. Ultrasonography images were collected under standard conditions (environmental temperature was 20–23 ◦C and the patient remained in a lying position for at least 10 min before examination). Acquired images were exported into a dedicated database and were evaluated using specific image-analysis software to assess epidermal and dermal thickness (μM) and lesional echogenicity.

RCM investigation on the same sides of the forearm was achieved with a Vivascope 1500® microscope (MAVIG GmbH, Lucid Technologies, Henrietta, NY, USA) to visualize *in vivo* the horizontal optical skin sectioning at cellular-level resolution (lateral resolution = 0.5–1 μM, axial resolution = 3–5 μM) from the epidermis to the papillary dermis (200–250 μM in depth). The system uses a laser source with a wavelength of 830 nm and a power <35 mW at the tissue level. The microscope objective is attached to the skin through an adhesive ring to diminish motion artefacts during investigation. Water was used between the adhesive window and the skin, and ultrasound gel (Aquasonic 100 Gel; Parker Laboratories Inc., Fairfield, NJ, USA) was used between the adhesive window and the lens as detailed in [35]. VivaScan 7.0, Viva StackTM and Viva BlockTM software (Lucid Technologies, Henrietta, NY, USA) was employed to acquire blocks of 4 × 4 mm horizontal optical sections, obtained from 64 individual horizontal optical sections (500 × 500 μm images). The system saves images in bitmap format with digital resolution of 1000 × 1000 pixels and 256 levels of grey.

#### **3. Results**

#### *3.1. Clinical Findings*

The proband (LOVD ID AN\_006205) was an Italian 53-year-old woman, born from healthy second-cousins parents, and had two healthy brothers. Clinical history was remarkable for premature birth at 30 weeks (height 44 cm, weight 1.2 kg) associated with perinatal respiratory distress. Neonatal severe hypotonia and delayed motor development, i.e., delays in walking (she took her first steps at four years of age) and acquisition of fine motor skills, were also reported. Medical history further included propensity to develop ecchymoses either spontaneously or upon minimal trauma often occurring for motor clumsiness, surgically treated umbilical hernia in infancy, myopia and astigmatism since childhood, and complete dental loss due to unspecified periodontitis at 14 years old. At age 18, a clinical diagnosis of unspecified EDS was given for gJHM, skin hyperextensibility, delayed wound healing, and easy bruising; genetic analyses were not performed. The patient suffered from recurrent dislocations of knees and occasionally of shoulders and elbows since the age of 10; the objective patellar instability was surgically treated by capsuloplasty and transposition of the insertion of the common patellar tendon by tibial tuberosity transplantation followed by skin plastic surgery at the age of 29 leading to a wide atrophic post-surgical scar (Figure 1A). At 21 and 23 years old, respectively, she underwent bilateral saphenectomy for symptomatic varicosities with pain, fatigability, heaviness, and recurrent superficial thrombophlebitis and surgical removal of nodules on vocal cords. At age 41, the patient was subjected to operative treatment of rotator cuff disease in the setting of weakness and substantial functional disability. Since age 42, she suffered from Achilles tendinopathy with pain and stiffness, especially at the back of the ankle, treated with on-demand NSAIDs use, conservative physical therapy, and orthotic insoles for severe pes planus. At 43 years of age, metatarsal osteotomy on the 3rd toe of the right foot for metatarsalgia and aggravating Achilles tendinopathy was performed. In the same period, she developed disabling bilateral gonarthrosis, treated with arthroscopic abrasion, epitrochleitis, and subacromial shoulder impingement associated with night pain. Hypotrophy of the scapular girdle and weak osteotendinous reflexes were observed at age 50, when she also experienced the dislocation of the left ankle with soft tissue effusion without reabsorption.

On examination, at 52 years of age, she presented with a height of 150 cm (genetic target 157 cm, arm span/height ratio 1.03, normal value <1.05), a weight of 52 kg, hyperextensible, soft, doughy, fragile and redundant skin, with an old-aging appearance of face and extremities, and multiple atrophic papyraceous scars, especially on knees, defective wound healing, easy bruising, spheroids on the elbows, and BS 5/9 (Figure 1A). She also showed multiple papules with some follicular prominence that looked like a diffuse poikiloderma of Civatte (PoC-like dermatitis) more pronounced in photo-exposed sites, androgenetic alopecia, high palate, elongated uvula, scoliotic attitude, mobile patellae and flat feet (even though surgical intervention and orthotics, respectively), hallux valgus, bilateral piezogenic papules, peripheral artery disease (i.e., intermittent claudication, peripheral cyanosis, and cold skin), and varicose veins (Figure 1A). The patient reported persistent lumbar back pain and sporadic pain of hips, knees, left ankle, elbows, shoulders, and feet. Multidimensional fatigue inventory (MFI) questionnaire was suggestive for chronic fatigue (total score 69, higher score in the questions investigating physical fatigue). Cognitive development and mentation were normal. Heart ultrasound detected normal cardiac/valve morphology and function. Dual-energy X-ray absorptiometry (DXA) disclosed femoral osteopenia (T-score left femoral neck −1.5 SD, T-score total hip −1.6 SD); lumbar BMD was normal (T score −0.9 DS) in the presence of marked degenerative arthritis. Nevertheless, we found a mild dorsal vertebral deformity (T10) in the presence of a low TBS value (1.23). The patient also presented mild scoliosis and lumbar spine rectilinization. Due to hypovitaminosis D, Cholecalciferol 50,000 UI monthly was commenced. Other bone metabolism blood and urinary samples and markers of bone remodeling were normal.

**Figure 1.** Clinical and molecular findings of the patient. (**A**) Old-aging appearance of face and androgenetic alopecia (a,b); laxity of the thumb (c), laxity of the fifth finger (d); hyperextensible skin in different body areas: neck (e), elbow (f), dorsum of the hand (g) and forearm (h); flat feet and piezogenic papules (i); subcutaneous spheroid on elbow (j), diffuse PoC-like dermatitis and easy bruising (k); skin redundancy, atrophic papyraceous scars on knees, postsurgical enlarged scar after right knee capsuloplasty and skin plastic surgery (l). (**B**) Sequence chromatograms showing the position of the c.1925T>C p.(Leu642Pro) variant (arrow) identified in homozygosity in exon 16 of the *AEBP1* gene (seq. Ref.: NM\_001129.4, NP\_001120.3) and *in silico* prediction of the pathogenicity of the p.(Leu642Pro) missense substitution by using 13 different algorithms [20–32].

#### *3.2. Molecular Findings*

The patient's phenotype was suggestive for cEDS, since she fulfilled both major (skin hyperextensibility plus atrophic scarring and gJHM) and 6 minor criteria according to the 2017 EDS nosology, i.e., easy bruising, soft, doughy skin, skin fragility, subcutaneous spheroids, a history of hernia, and JHM complications. Therefore, after written informed consent was obtained, we performed Sanger sequencing of *COL5A1*, *COL5A2*, and of exon 14 of *COL1A1* (p.Arg312Cys), integrated by MLPA analysis of *COL5A1*, which did not identify any pathogenic variant. Although negative molecular testing, a clinical diagnosis of cEDS was maintained, since the other EDS types in differential diagnosis with cEDS (including periodontal EDS) were excluded clinically. Following the discovery of *AEBP1* biallelic variants [15,16], Sanger sequencing of this gene was achieved, which revealed the homozygosity for the rare c.1925T>C p.(Leu642Pro) variant in exon 16 (Figure 1B), leading to the substitution of a highly conserved leucine residue with a proline at position 642 within the metallocarboxypeptidase-like domain of the protein. This variant has been observed in 3 individuals in GnomAD (rs753531562, 3/282140, no homozygotes, total MAF: C = 0.00001063). Its putative pathogenicity was estimated through an array of 13 different *in silico* prediction algorithms that agreed to define p.(Leu642Pro) as high impacting variant. Given that the variant is located in a critical and well-established functional domain without benign variation, the extremely low frequency in publicly available population databases, the multiple lines of computational evidence supporting a deleterious

effect on the gene product, and the patient's phenotype highly suggestive for a disease with a single genetic etiology, the p.(Leu642Pro) missense variant is classified as likely pathogenic (class 4) according to the guidelines of the ACMG. Samples of the healthy parents or brother were not available for molecular analyses.

#### *3.3. Instrumental Findings on Patient's Skin*

In order to investigate the skin by a non-invasive approach, HF-USG and *in vivo* RCM were performed on selected skin areas, i.e., dorsal and volar side of the forearm, showing clinically significant differences between our patient and 10 healthy individuals (Table 1 and Figure 2). Digital 50-MHz ultrasonography scanning demonstrated an increase in epidermal entrance echo (highly echogenic band produced by the differences of the acoustic impedance between gel and skin) corresponding to increased epidermal thickness, but a decrease in dermal thickness compared to control skin of age- and gender-matched healthy individuals with the same skin phototype II and similar sun exposure history. The patient's epidermis (dorsal thickness = 172 μM; volar thickness = 141 μM) was thicker than that of healthy controls (dorsal thickness (mean ± standard deviation, SD) = 121 ± 22 μM; volar thickness (mean ± SD) = 102 ±12 μM), likely due to the multiple and diffuse papules (Table 1). The increased thickness was more evident on the dorsal side of the forearm that is chronically more photoexposed compared to the volar side. Contrariwise, the patient's dermis appeared thinner (dorsal dermal thickness = 570 μM; volar epidermal thickness = 289 μM) compared to healthy controls (dorsal dermal thickness (mean ± SD) = 1108 ± 320 μM; volar epidermal thickness (mean ± SD) = 983 ± 205 μM) (Table 1). Moreover, the considerable hypoechogenicity of the dermal layer suggests disruption of collagen fibers and accumulation of elastotic material that is typical of chronological and photoinduced skin aging (Figure 2A). This ultrastructural pattern is known as subepidermal low echogenic band (SLEB) and derives from skin elastosis and accumulation of glycosaminoglycans that have increased water-binding capacity [33]. *In vivo* RCM investigation demonstrated loss of the typical honey-comb pattern (corresponding to alteration of epidermal thickness), irregularity of the dermal-epidermal junction and the disarray of the dermis, which was characterized by coarse and fragmented collagen fibers both on the dorsal and volar side of patient's forearm (Figure 2B). These alterations are independent of sun exposure, given that they are present both on dorsal and volar side of the forearm, suggesting a pronounced and diffuse skin aging due to *AEBP1*-defect.


**Table 1.** Epidermal and dermal thickness of patient' s forearm evaluated by high-frequency ultrasonography (HF-USG) compared to 10 healthy individuals.

#### **4. Discussion**

Recently, taking advantage from NGS, a new, autosomal recessive type of EDS has been discovered due to variants in the *AEBP1* gene. This EDS type is very rare and, so far, found in only seven individuals (including the present patient) from four unrelated families (Table 2). The International Consortium on EDS and Related Disorders has not yet classified and named this type, but in OMIM it is labeled as classical-like EDS type 2 (MIM #618000). Indeed, the few patients reported hitherto (Table 2) share many similarities representative of the classical type as much as they all fulfill the cEDS diagnostic criteria of the 2017 nosology [1,2] for the presence of the pathognomonic cutaneous involvement, i.e., soft, doughy and very hyperextensible skin, delayed wound healing with abnormal atrophic scarring, JHM, and other minor criteria such as easy bruising, subcutaneous spheroids (observed only in our patient), and JHM complications such as dislocations/subluxations (shoulders, knees, hips, ankles, elbows, clavicula, wrist, mandibular and distal radioulnar joints, in some cases requiring surgical treatment), sprain, pain, and flexible flatfoot (Table 2)

**Figure 2.** Instrumental findings on patient's skin. (**A**) Ultrasonography (50 MHz) images of the forearm skin from the patient and a representative age- and gender-matched healthy individual (control). E, epidermis, D, dermis, S, subcutaneous adipose tissue (depth of imaging: 4 mm). Disorganization of collagen fibers and elastosis in patient's skin appears as a significant thinning and hypoechogenicity of the dermal layer both on dorsal (left) and volar side (right) compared to control (**B**) Reflectance confocal microscopy images of the forearm skin from patient and control (magnification: 500 × 500 μm). Epidermis: typical honey-comb pattern on dorsal (left) and volar (right) side in healthy skin are not detectable in patient's skin. Dermal-epidermal junction: regular edge papillae [rings of basal keratinocytes surrounding dark circular structures corresponding to dermal papillae (\*)] on dorsal and volar side of control skin are reduced both in number and definition in patient's skin. Dermis: Irregular and fragmented collagen fibers that appear bright and coarse on both dorsal and volar side of patient's skin compared to control. An increased brightness of all skin structures, corresponding to chronological and photoinduced skin aging, is present on the dorsal side of healthy skin but not on the volar side that usually is not photoexposed, whereas in the patient this pronounced skin aging is present at both sides.


#### *Genes* **2019** , *10*, 135


**Table 2.** *Cont.*

*Genes* **2019**, *10*, 135

Consistent with the multisystemic presentation of EDS in general, there are also variable features including congenital hip dislocation, hypotonia, delayed motor development, acrogeria, prominent superficial veins in the chest region, hernias, dental anomalies, gastrointestinal (bowel rupture) and vascular complications (mitral valve prolapse, aortic root dilation needing surgery), early-onset varicose veins, and several skeletal anomalies (Table 2). In particular, bone involvement seems a common feature of classical-like EDS type 2 with osteopenia/osteoporosis affecting hips, femurs, and spine that are present, at variable degree, in all of the patients reported so far, with the exception of the two siblings, reported by Hebebrand and coworkers [17], who were not tested for osteopenia. In addition, degenerative arthritis, (kypho)scoliosis, arachnodactyly, positive wrist and thumb signs, mild pectus excavatum, T10 vertebral deformity (our patient), narrowing of the interpedicular distance of the lumbar spine, shortened and squared iliac bones, and remodeling of long bones of the lower extremities are also encountered (Table 2). In addition, all subjects have severe foot deformities including bilateral pes planus, hammertoes, hallux valgus, hindfoot deformity, and sandal gap, which are observed in more than a few other EDS subtypes as well [1]. Although in cEDS patients a variable degree of low bone mineral density and a high prevalence of radiological vertebral fractures were reported [7,36], *AEBP1*-related EDS seems to display a more severe bone involvement that could potentially facilitate the differential with cEDS. Nevertheless, considering the limited number of individuals with *AEBP1* defect known so far, a larger cohort of patients is needed to confirm this preliminary observation.

The adipocyte enhancer binding protein 1 gene (*AEBP1)* encodes a 1158-amino acid secreted aortic carboxypeptidase-like protein (ACLP) composed of an N-terminal signal sequence, a charged lysine, proline, and glutamic acid-rich domain, a collagen-binding discoidin domain and a metallocarboxypeptidase (MCP)-like domain [37,38]. This latter domain is inactive toward standard MCP substrates, as it lacks several critical active sites and substrate-binding residues that are necessary for activity [37,38]. Indeed, ACLP acts as an extracellular matrix (ECM)-binding protein rather than as active MCPs that shows similar embryonic expression pattern as other ECM proteins and is found at high levels particularly in collagen-rich tissues comprising the dermal layer of the skin, the medial layer of blood vessels, the basement membrane of the lung, and the periosteum. Consistently, ACLP plays fundamental roles in both embryonic development and adult tissue homeostasis, particularly in repair processes [38–43]. Indeed, *AEBP1* knock-out mice show ventral wall defects, develop spontaneous skin ulcerations, and have significantly delayed healing of dermal punch wounds [38]. This cutaneous phenotype is consistent with the defective wound healing and abnormal scar formation observed in individuals with *AEBP1* defects and suggest that ACLP has a crucial role in damage sensing and ECM remodeling following injury by regulating fibroblast proliferation and mesenchymal stem cell differentiation into collagen-producing cells [42,43]. Blackburne and coworkers demonstrated that ACLP also binds collagens type I, III, and V and is able to promote the polymerization of collagen type I in vitro [16]. In line with these findings, the ultrastructural study performed by the same authors on a patient's skin biopsy revealed reduced dermal collagen and irregular disrupted collagen fibers, as well as our HF-USG and RCM investigations that disclosed abnormal collagen fibers deposition together with a reduced dermal thickness. Moreover, we recognized an increase of the epidermal thickness likely correlating with the diffuse PoC-like dermatitis, which is probably not related to classical-like EDS type 2. The use of these non-invasive diagnostic techniques may be promising for the investigation of the qualitative and quantitative cutaneous alterations, but further studies including electron microscopy on skin as golden standard of reference on large cohorts of patients are warranted. In our case, we did not perform skin biopsy because of the patient's will due to psychological reluctance for her important skin fragility with delayed wound healing.

The *AEBP1* variants discovered before our patient's characterization were all loss-of-function (LOF) mutations (Table 2) and included compound heterozygous variants [(c.1470del; p.Asn490\_Met495delins40) and (c.1743C>A; p.Cys581\*] in the first individual (P1); a homozygous variant (c.1320\_1326del; p.Arg440Serfs\*3) in the second individual (P2); a homozygous splice

variant leading to skipping of the last 22 bp of exon 13 (c.1630+1G>A) in the two siblings from the third family (P3, P4), and a homozygous nonsense variant (c.917dup; p.Tyr306\*) in the two siblings from the fourth family (P5, P6). Hebebrand and coworkers performed the analysis of all *AEBP1* LOF variants reported in multiple databases showing that these are distributed throughout the protein and by using conservative criteria for pathogenic LOF variants (nonsense, frameshift, canonical splice sites, or initiation-codon) these authors estimated a carrier frequency of 1/829 for the gnomAD database. The analysis of CADD scores for all possible missense variants showed a higher predicted deleteriousness for positions close to the discoidin and the MCP-like domains, whereas the unstructured N-/C-terminal parts showed lower scores. The high deleteriousness scores observed for missense variants within these domains cite evidence in support of additional mutational mechanisms leading to aberrant function and the authors thus argued that the relatively low estimated carrier frequencies could be significantly higher if missense variants contribute to a comparable fraction of disease variants [17]. The present c.1925T>C; p.(Leu642Pro) homozygous variant disclosed within the MCP-like domain of the protein corroborates this hypothesis, since it represents the first likely pathogenic *AEBP1* missense substitution (ACMG class 4) associated with classical-like EDS type 2. The variant is predicted *in silico* to affect the tertiary structure of the protein by disrupting an α-helix located in a highly conserved domain, thus likely interfering with its function in terms of impaired partner binding capability. Nevertheless, a definite proof of variant' s causality is lacking, since the effective functional consequences on the ECM organization, particularly of collagens, and on the other not yet well-defined roles of the ACLP protein were not studied, because the patient refused skin biopsy.

#### **5. Conclusions**

Our findings expand the knowledge of the clinical phenotype of this recently defined autosomalrecessive EDS subtype, provide the first evidence that missense variants contribute to the allelic repertoire of *AEBP1,* and suggest that in the diagnostic process of a cEDS patient this gene should be investigated when a recessive inheritance is compatible and no causal variant is identified in the other cEDS genes. Further reports are needed to better characterize the *AEBP1*-related phenotype, define specific clinical criteria that might facilitate the differential with the other EDS forms, delineate genotype-phenotype correlations, and collect natural history data for prognostication. Finally, ACLP function needs to be explored more in-depth to provide insights into molecular mechanisms involved in the pathophysiology of *AEBP1*-related EDS that may represent a starting point for identifying potential therapeutic options.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/2/135/s1, Table S1. Primers.

**Author Contributions:** M.C. and M.R. conceived the study. M.V. and M.C. performed the clinical diagnosis of the patient, genetic counselling and follow-up; M.V. performed skin evaluations; L.P. and A.M.F. investigated bone health parameters; M.R. and V.C. carried out the molecular analyses; M.R. and N.C. researched the literature; M.R., A.M.F., and M.V. prepared the manuscript; M.C. edited and coordinated the manuscript. All authors discussed, read, and approved the manuscript.

**Funding:** No funding was active on this project.

**Acknowledgments:** The authors want to thank the patient for her cooperation during the diagnostic process and the Fazzo Cusan family for its generous 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/).

### *Article* **Clinical and Molecular Characterization of Classical-Like Ehlers-Danlos Syndrome Due to a Novel** *TNXB* **Variant**

**Daisy Rymen 1,\*,**†**, Marco Ritelli 2,**†**, Nicoletta Zoppi 2, Valeria Cinquina 2, Cecilia Giunta 1, Marianne Rohrbach <sup>1</sup> and Marina Colombi <sup>2</sup>**


Received: 25 September 2019; Accepted: 23 October 2019; Published: 25 October 2019

**Abstract:** The Ehlers-Danlos syndromes (EDS) constitute a clinically and genetically heterogeneous group of connective tissue disorders. Tenascin X (TNX) deficiency is a rare type of EDS, defined as classical-like EDS (clEDS), since it phenotypically resembles the classical form of EDS, though lacking atrophic scarring. Although most patients display a well-defined phenotype, the diagnosis of TNX-deficiency is often delayed or overlooked. Here, we described an additional patient with clEDS due to a homozygous *null*-mutation in the *TNXB* gene. A review of the literature was performed, summarizing the most important and distinctive clinical signs of this disorder. Characterization of the cellular phenotype demonstrated a distinct organization of the extracellular matrix (ECM), whereby clEDS distinguishes itself from most other EDS subtypes by normal deposition of fibronectin in the ECM and a normal organization of the α5β1 integrin.

**Keywords:** Tenascin X; TNXB; Ehlers-Danlos syndrome; EDS; connective tissue; collagen

#### **1. Introduction**

The Ehlers-Danlos syndromes (EDS) constitute a clinically and genetically heterogeneous group of connective tissue disorders. Patients present with joint hypermobility, skin hyperextensibility and tissue fragility, giving rise to easy bruising and atrophic scarring. Although most EDS are caused by mutations in genes coding for the fibrillar collagens or collagen-modifying enzymes, over the last two decades, several disorders due to defects in other components of the extracellular matrix (ECM) have been delineated [1].

Tenascins comprise a family of glycoproteins, which modulate the adhesion of cells to their ECM. Tenascin X (TNX) is ubiquitously expressed, but the highest levels are found in muscle and loose connective tissue [2]. TNX is thought to regulate fibril spacing by direct binding to the distinct collagen fibrils in the ECM or by indirect binding via decorin [3]. Additional roles for TNX in elastic fiber remodeling and regulating the expression of certain ECM components, e.g., collagen VI, proteoglycans and matrix metalloproteases have been suggested [4–7].

The clinical relevance of TNX-deficiency was first proposed in 1997 by the identification of a patient presenting both congenital adrenal hyperplasia and an EDS-like phenotype. Genetic analysis revealed a contiguous deletion of the partially overlapping genes *CYP21B* and *TNXB* on chromosome 6. Although a TNX *null*-phenotype was verified both on the protein and on the mRNA level, a mutation on the second allele could not be identified [8]. In 2001, Schalkwijk et al. demonstrated that isolated TNX-deficiency resulted in an autosomal recessive form of EDS, resembling the classical type, however lacking atrophic scarring [9]. In 2017, TNX-deficiency was officially classified as "classical-like EDS" (clEDS), with generalized joint hypermobility, hyperextensible, soft and/or velvety skin without atrophic scarring and easy bruising being the typical clinical hallmarks of the disorder [1].

In 2002, Mao et al. developed a TNX-deficient mouse model, mimicking the EDS phenotype. Indeed, as observed in humans, *Tnxb-*/*-* mice were morphologically normal at birth, but displayed progressive hyperextensibility of the skin over time. The group showed that the phenotype did not relate to aberrant collagen fibrillogenesis, but was rather due to altered deposition, and therefore, reduced density of collagen in the ECM [10].

Although most TNX-deficient patients display a well-defined clinical phenotype, the diagnosis is often delayed or overlooked. The former is attributed to the molecular analysis of the *TNXB* gene being complicated by the presence of a highly homologous pseudogene and to the fact that the measurement of TNX in serum is not widely available [11]. The latter is mainly caused by poor clinical awareness, which unfortunately applies to many rare disorders.

Here, we reported on an additional patient with clEDS and a novel homozygous disease-causing variant in *TNXB* to further elaborate the clinical phenotype. Furthermore, we reviewed the clinical features of the clEDS patients described to date, in order to create a well-defined description of the phenotype and increase clinical awareness.

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

#### *2.1. Ethical Compliance*

This study is in accordance with the Helsinki declaration and its following modifications. Ethics approval has been granted (KEK Ref.-Nr. 2014-0300 and Nr. 2019-00811) in the presence of a signed informed consent of the patient for genetic testing, skin biopsy and the publication of clinical pictures. The patient was evaluated at the University Children's Hospital of Zürich. Targeted next-generation sequencing (NGS) panel for 101 connective tissue disorders (Supplementary Table S1) was performed at the Institute of Medical Genetics of the University of Zürich. *TNXB* mutational screening by Sanger sequencing and Multiplex Ligation-dependent Probe Amplification (MLPA) was achieved at the Division of Biology and Genetics, Department of Molecular and Translational Medicine, of the University of Brescia.

#### *2.2. Cell Culture*

As part of the diagnostic workup of EDS, a punch biopsy of the patient's skin for establishing fibroblast cultures for collagen biochemical analysis was previously obtained. The biological material was stored in the Biobank of the Division of Metabolism at the Children's Hospital Zürich. Fibroblasts from the patient and from sex- and age-matched healthy individuals were routinely maintained at 37 ◦C in a 5% CO2 atmosphere in Earle's Modified Eagle Medium (MEM) supplemented with 2 mM L-glutamine, 10% FBS, 100 μg/ml penicillin and streptomycin (Life Technologies, Carlsbad, CA, USA). Fibroblasts were expanded until full confluency and then harvested by 0.25% trypsin/0.02% EDTA treatment at the same passage number.

#### *2.3. Molecular Analysis*

Mutational screening was performed on genomic DNA purified from peripheral blood leukocytes using standard procedures. In particular, all exons and their intron-flanking regions of the *TNXB* gene (NM\_0019105.7, NP\_061978.6) were PCR amplified with the GoTaq Ready Mix 2X (Promega, Madison, WI, USA) by using optimized genomic primers that were analyzed for the absence of known variants using the GnomAD database (https://gnomad.broadinstitute.org/). For the pseudogene-homolog region

(exons 32–44), Sanger sequencing was performed by nested PCR, using a *TNXB*-specific long-range PCR product encompassing the 3'-end of *TNXB* as a template (for details on primer sequences and PCR conditions see Supplementary Table S2). PCR products were purified with ExoSAP-IT (USB Corporation, Cleveland, OH, USA), followed by bidirectional sequencing with the BigDye Terminator v1.1 Cycle Sequencing kit on an ABI3130XL Genetic Analyzer (Thermo Fisher Scientific, South San Francisco, CA, USA). 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). Deletion/duplication analysis of *TNXB* was performed using the MLPA assay P155, according to manufactures' instructions (MRC-Holland, Amsterdam, the Netherlands).

#### *2.4. RNA Extraction and Quantitative Real-Time PCR*

Total RNA was purified from skin fibroblasts of the patient and 3 healthy individuals using the Qiagen RNeasy kit, according to the manufacturer's instructions (Qiagen, Hilden, Germany). RNA quality control was performed on an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Relative expression levels of the *TNXB* transcript were analyzed by quantitative real-time PCR (qPCR). Of the total RNA, 3 μg was reverse-transcribed with random primers by a standard procedure. qPCR reactions were performed in triplicate with the SYBR Green qPCR Master Mix (Thermo Fisher Scientific, South San Francisco, CA, USA), 10 ng of cDNA, and with 10 μM of each primer set by using the ABI PRISM 7500 Real-Time PCR System with standard thermal cycling conditions. The *HPRT*, *GAPDH* and *ATP5B* reference genes were also amplified for normalization of cDNA loading. Relative mRNA expression levels were normalized to the geometric mean of these reference genes and analyzed using the 2−(ΔΔCt) equation. Amplification plots, dissociation curves and threshold cycle values were generated by ABI Sequence detection system software version 1.3.1. Statistical analyses were performed with the GraphPad Prism software (San Diego, CA, USA). The results were expressed as the mean values of relative quantification ± Standard Error of the Mean (SEM). Statistical significance between groups was determined using one-way ANOVA. *P*-values were corrected for multiple testing using the Tukey's method.

#### *2.5. Collagen Biochemical, Ultrastructural and Immunofluorescence Microscopy Studies*

Collagen steady state analysis in patient's cultured fibroblasts was conducted prior to genetic analysis to assess possible anomalies in collagen biosynthesis and secretion, as previously described [12]. Briefly, radioactively labeled and pepsinized procollagens from the patient and one healthy individual were separated on a 5% Sodium Dodecyl Sulphate Polyacrylamide Gel Electrophoresis (SDS-PAGE) and visualized by autoradiography.

To analyze ECM organization of collagen type I (COLLI), collagen type III (COLLIII), collagen type V (COLLV) and fibronectin (FN), skin fibroblasts derived either from the patient or an unrelated control individual were grown for 72 hours in the presence of 50 μM ascorbic acid (for COLLI only), refed after 24 hours and immunoreacted as previously reported [13–15]. In brief, cold methanol fixed fibroblasts were immunoreacted with 1:100 anti-COLLI, anti-COLLIII, anti-COLLV (Millipore-Chemicon Int., USA), and anti-FN (Sigma Chemicals, St. Louis, MO, USA) antibodies (Ab). The ECM of TNX was investigated on methanol-fixed cells immunoreacting with 2 μg/ml anti-TNX Ab (Santa Cruz Biotec. Inc., USA). For the analyses of the α2β1, α5β1 and αvβ3 integrin receptors, cells were fixed in 3% paraformaldehyde (PFA)/60 mM sucrose and permeabilized with 0.5% Triton X-100 as previously reported [13,14,16]. In particular, fibroblasts were reacted for 1 hour at room temperature with 4 μg/ml anti-α5β1 (clone JBS5), anti-α2β1 (clone BHA.2) and anti-αvβ3 (clone LM609) integrin monoclonal Abs (Millipore-Chemicon Int., USA). The cells were then incubated for 1 h with Alexa Fluor 488 anti-rabbit and Alexa Fluor 555 anti-mouse Ab (Thermo Fisher Scientific, South San Francisco, CA, USA), or with rhodamine-conjugated anti-goat IgG (Calbiochem-Merck, Germany). Immunofluorescence (IF) signals were acquired by a black-and-white CCD TV camera (SensiCam-PCO Computer Optics GmbH,

Germany) mounted on a Zeiss fluorescence Axiovert microscope and digitalized by Image Pro Plus software (Media Cybernetics, Silver Spring, MD, USA).

#### **3. Results**

#### *3.1. Case Report*

The patient was referred to our center at the age of 41 years because of a suspected connective tissue disorder. She was the first child of consanguineous parents of Swiss origin. The family history was unremarkable. The patient displayed a congenital dislocation of the left hip. Due to recurrent dislocations, an arthrodesis was performed at the age of 2 years. Immediately after surgery, a fracture of the left femur occurred, requiring osteosynthesis. Over the years, multiple joint dislocations ensued, mainly of the shoulders. During childhood, the patient developed a progressive kyphoscoliosis, which was left untreated. At present, the patient displays a right sided gibbus deformity (Figure 1). Upon radiography of the spine, a Cobb angle of 22, 38 and 15 degrees was measured at the cervical, thoracic and lumbar level, respectively.

**Figure 1.** X-ray images of the vertebral column of the patient at the age of 41 years. (**a**) Side view demonstrating the pronounced kyphosis; (**b**) close-up of the pronounced cervical and thoracic scoliosis; (**c**) close-up of the lumbar scoliosis.

In addition, the patient presented a progressive arthrosis of the knee and degenerative changes of the spine. Bone densitometry at the age of 41 years was normal. Generalized joint hypermobility was absent (Beighton score 3/9). The patient reported mild muscular weakness. Computed tomography of the abdomen demonstrated fatty degeneration of the abdominal, gluteal and pelvic floor musculature. The patient was noted to have a soft, dough-like skin texture associated with skin hyperextensibility. In addition, the skin was thin and translucent with multiple varicose veins and edema of the lower extremities (Figure 2c). Redundant and sagging skin was found on both knees (Figure 2a). Striae were already present at young age. Feet and hands were short and broad, with the presence of brachydactyly (Figure 2d,g). Acrogeria of the hands was noticed, with an increased amount of palmar skin creases (Figure 2e). Several subcutaneous nodules were present at the interphalangeal joints of the fingers. Piezogenic papules were present on both feet (Figure 2f). While standing upright, the subcutaneous tissue of the feet was pushed aside, leading to neurogenic pain and the inability to walk barefoot or to go longer distances (Figure 2f). The patient reported large hematomas to occur after minimal trauma. Atrophic scarring was not present (Figure 2b). However, wound healing was clearly delayed with wound dehiscence after surgery. The patient presented with an inguinal and umbilical hernia at the age of 27 and 36 years, respectively, requiring surgical correction. The patient mothered three

children, all born through C-section. A urogenital or rectal prolapse did not occur. The patient did, however, suffer from hemorrhoids. Colonoscopy at the age of 40 years demonstrated asymptomatic diverticulosis of the sigmoid colon. The patient presented with one episode of ileitis of unknown origin. In addition, the patient had two episodes of dactylitis and suffers from recurrent aphthous stomatitis. No biochemical indication of an underlying inflammatory disease was found. At the age of 42 years, a spontaneous perforation of the small bowel occurred, necessitating surgery and resection of the involved bowel loop. The intestinal specimen obtained during surgery was reported to be very fragile. An echocardiography at the age of 41 years demonstrated an intact mitral valve and normal dimensions of the aorta. The patient was known with a mild myopia. Hearing was normal to date.

**Figure 2.** Clinical features of the patient. (**a**) Redundant and sagging skin on both knees; (**b**) widened scar on the left knee after surgery. No atrophic features are present; (**c**) chronic ankle edema in the absence of cardiac failure; (**d**) plane flat feet with broad forefeet and short digits; (**e**) increased palmar skin creases; (**f**) piezogenic papules; (**g**) short and broad acrogeric hands. Note the hyperextensibility of the thumb.

#### *3.2. Molecular and Biochemical Findings*

#### 3.2.1. Molecular Analysis

Previous molecular investigations of the patient, all with negative results, included molecular genetic analysis of *COL1A1* and *COL1A2* on cDNA for the suspicion of arthrochalasis EDS, and an NGS panel for 101 connective tissue disorders. Given that the *TNXB* gene was not present in the NGS panel and considering the clinical presentation of the patient that was suggestive for clEDS, we sequenced *TNXB* by traditional Sanger method. Sequencing of all exons and exon-intron boundaries of *TNXB* (NM\_0019105.7, NP\_061978.6) revealed the homozygous c.5362del, p.(Thr1788Profs\*100) variant in exon 15 (Figure 3a), which was absent in all public variant databases. Hence, the novel pathogenic *TNXB* variant (identifier #0000591583) was submitted to the Leiden Open Variation Database (LOVD, https://databases.lovd.nl/shared/genes/TNXB).qPCR analysis on cDNA obtained from patient's dermal fibroblast showed that the c.5362del transcript, which leads to frameshift and formation of a premature termination codon (PTC) p.(Thr1788Profs\*100), represents a null-allele undergoing nonsense-mediated mRNA decay (Figure 3b). Consistently, IF analysis on the patient's fibroblasts with a specific Ab against TNX demonstrated a complete absence of the protein in the ECM (Figure 3c).

**Figure 3.** The *TNXB* c.5362del pathogenic variant causes nonsense-mediated mRNA decay and absence of the tenascin X protein in the ECM. (**a**) Sequence chromatograms showing the position of the novel c.5362del; p.(Thr1788Profs\*100) variant (arrow) identified in homozygosity in exon 15 of the *TNXB* gene by Sanger sequencing; (**b**) qPCR analysis on cDNA obtained from patient's dermal fibroblasts showed that the transcript with the p.(Thr1788Profs\*100) frameshift variant undergoes nonsense-mediated mRNA decay. The relative quantification (RQ) of the *TNXB* transcript levels was determined with the 2−(ΔΔCt) method normalized with the geometric mean of the *HPRT*, *GAPDH* and *ATP5B* reference genes. Bars represent the mean ratio of target gene expression in patients' fibroblasts compared to three unrelated healthy individuals. qPCR was performed in triplicate, and the results are expressed as means ± SEM. The relative mRNA level of *TNXB* in the patient versus controls (about 142-fold decrease) is expressed as Log10 transformed value. Statistical significance (\*\*\* P < 0.001) was calculated with one-way ANOVA and the Tukey post hoc test. (**c**) IF analysis on patient's skin fibroblasts with a specific antibody against TNX showing the absence of the protein in the ECM compared to control cells. Scale bar: 10 μm.

#### 3.2.2. Collagen Biochemical and Immunofluorescent Analysis

To assess the effects on collagen biosynthesis and secretion, we performed collagen steady-state and pulse-chase analyses in dermal fibroblasts from the patient. Normal relative proportions of COLLI, COLLIII and COLLV were present in the cell lysate and in the medium (Supplementary Figure S1). There were no indications for altered modification of the different collagens.

To investigate the effect of TNX-deficiency on COLLI-, COLLIII-, COLLV- and FN-ECM organization and on the expression of their integrin receptors, dermal fibroblasts from the patient and one healthy individual were investigated by IF (Figure 4). Control fibroblasts organized a reticular ECM of COLLIII and COLLV, rare fibrils of COLLI, an abundant FN-ECM and expression of the canonical α2β1 and α5β1 integrins, whereas patient's fibroblasts showed a lack of organized COLLI, COLLIII, and COLLV in the ECM, though proteins were present at different levels inside the cells. The disorganization of the COLLs-ECM is associated with a strong reduction of their canonical α2β1 integrin receptor (Figure 4a). Contrarily, the FN-ECM organization and the expression of its canonical integrin receptor α5β1 were not affected in TNX-deficient cells. Therefore, the alternative FN-receptor αvβ3 was almost undetectable both in control and patient cells (Figure 4b).

**Figure 4.** TNX-deficiency leads to the disarray of collagens type I, type III and type V in the ECM together with the reduction of the α2β1 integrin. (**a**) IF analysis with specific Ab demonstrates that TNX-deficient cells are characterized by the lack of organized COLLI, COLLIII and COLLV in the ECM, though proteins are present at different levels inside the cells. The disorganization of the COLLs-ECM is associated with a strong reduction of their canonical α2β1 integrin receptor. (**b**) TNX-deficient cells deposit FN in the ECM and express its canonical integrin receptor α5ß1 similarly to control fibroblasts. The alternative αvβ3 integrin is almost undetectable both in control and patients' cells. Scale bars: 10 μm.

#### **4. Discussion**

Most known EDS subtypes are autosomal dominant inherited disorders due to defects in genes coding for the fibrillar collagens or collagen-modifying enzymes. TNX-deficiency differs from the more frequent subtypes not only in its autosomal recessive inheritance pattern but also in the fact that it is caused by a defect in an ECM component other than collagen. The presumed role of TNX in the stabilization and maturation of the collagen- and elastin-ECM is reflected in the cellular and clinical phenotype of the patients described to date [2–7]. Similar to the data in *Tnxb-*/*-* mice, we could demonstrate that collagen biosynthesis and secretion were unaffected in our patient's fibroblasts, whereas a disarray of the COLLI-, COLLIII- and COLLV-ECM and a strong reduction of their canonical α2β1 integrin receptor was observed in vitro (Figure 4a) [10]. Unlike the typical EDS cellular phenotype, our patient's fibroblasts were characterized by a normal organization of the FN-ECM and therefore a normal expression of its canonical integrin receptor α5β1 (Figure 4b) [10,17].

Thus far, about 30 patients with complete TNX-deficiency have been described in literature [8,9, 18–26]. Most patients displayed either a complete lack of TNX in serum or biallelic mutations in *TNXB* leading to nonsense-mediated mRNA decay (Supplementary Table S3). As expected, those presenting missense mutations had milder or late-onset clinical manifestations [17,24].

As its name implies, TNX-deficiency or classical-like EDS (clEDS) phenotypically resembles the classical form of EDS (cEDS), with the triad of soft/velvety hyperextensible skin (20/20), generalized joint hypermobility (15/19) and a varying degree of tissue fragility (20/20) as its main clinical features (Table 1) [9]. Unlike cEDS, atrophic scarring is absent in patients with clEDS. However, about 50% of TNX-deficient patients present delayed wound healing with wound dehiscence upon suture removal and widened scars (Figure 2b). Data in *Tnxb-*/*-* mice suggest that, contrary to cEDS, abnormal wound healing in clEDS is not due to altered matrix deposition in the early phases of wound healing, but is rather caused by decreased ECM stabilization and maturation during the later stages. Indeed, contrary to the high levels of *TNXB* in normal skin, its expression during the early phase of wound healing is non-existing and only increases over time [27].

Whereas joint hypermobility seems to decrease with age, recurrent dislocations of the large joints (18/19) remain a problem over time and can be considered as the most frequent debilitating finding in the TNX-deficient patients described to date (Table 1).

Interestingly, several patients display deformities of the hands and feet, most of which can be ascribed to the underlying connective tissue weakness, such as pes planus, hallux valgus, piezogenic papules and acrogeric hands. Conversely, the presence of brachydactyly and broad hands and feet point to a role of TNX in development [28]. In addition, our patient presented chronic venous insufficiency, varicose veins and non-pitting ankle edema, which has thus far been observed in almost 30% of the TNX-deficient population (Table 1). The venous insufficiency and its consequences might be explained by the aberrant elastic fiber remodeling observed in TNX-deficient dermal fibroblasts and possible changes in endothelial cell proliferation [5,29,30]. Indeed, the interaction between TNX and vascular endothelial growth factor B (VEGF-B) is known to stimulate endothelial cell proliferation [30,31].

Bone is considered to be a target organ of osteogenesis imperfecta rather than EDS. However, premature osteopenia or osteoporosis have been published in various types of EDS and have been mainly attributed to abnormal COLLI fibrillogenesis and ECM deposition [32]. Although bone loss due to an increased number of multinucleated osteoclasts has been found in *Tnxb-*/*-* mice, thus far, indications for skeletal fragility in clEDS have not been reported [33]. Moreover, in our patient, bone densitometry at the age of 41 years was normal.

Over time, neuromuscular symptoms have been increasingly recognized in various types of EDS, ranging from muscle weakness to myalgia and easy fatigability [21]. Indeed, a normal composition of the ECM and intact innervation are important for adequate functioning of the muscle [34]. Considering the cellular phenotype of clEDS, neuromuscular involvement is not surprising. Collagen VI, which is deficient in two myopathies, i.e., Ullrich congenital muscular dystrophy and Bethlem myopathy, is downregulated, although with conserved organization of the ECM, in as well TNX-*null* fibroblasts and *Tnxb-*/*-* mice [7,34,35]. In addition, TNX is widely distributed in peripheral nerve and its deficiency causes axonal polyneuropathy in almost 40% of the patients (Table 1) [21]. Our patient had a low muscular mass on clinical examination and reported distal muscle weakness. Computed tomography of the abdomen demonstrated fatty degeneration of the muscles as an incidental finding, a well-known

phenomenon in myopathies of various origins. Interestingly, the neuromuscular phenotype seems to be mild or absent in children and progresses with age.


**Table 1.** Summary of clinical features of patients with biallelic *TNXB* variants.

Note: 19 out of 30 patients reported in literature were included. Patient 3 reported in Schalkwijk et al. [9], the index patient reported in Burch et al. [8], the three patients published in Chen et al. [25] were excluded because of concomitant congenital adrenal hyperplasia. The patient described in Mackenroth et al. [26] was excluded because of concomitant *COL1A1* mutation. Patient 4 and 5 reported in Schalkwijk et al. [9] were excluded because a lack of clinical data. The patient described in O'Connell et al. [24] and the two patients described by Lindor et al. [18] were excluded because no mutation analysis was performed.

Extra-articular symptoms in clEDS, such as gastrointestinal and cardiovascular complications, do not seem to be associated with certain mutations, but are rather related to age and have been reported after the 3rd–4th decade [18,22,23,25]. Gastrointestinal manifestations, although rare, can

lead to life-threatening situations. Spontaneous perforation of hollow organs is a complication mainly seen in patients with vascular EDS (vEDS) due to mutations in COLLIII and has only sporadically been described in other EDS subtypes. Compared to other non-vEDS, a higher prevalence seems to exist in clEDS, ranging from tracheal or esophageal rupture to spontaneous perforation of colon, diverticula or small intestine [18,20,22,23,25,36]. Therefore, patients and treating physicians should be aware of this risk in order to minimize the occurrence of iatrogenic perforation and to not delay the diagnosis and treatment of spontaneous events. Contrary to patients with vEDS, aortic root dilatation or aneurysms are not common in clEDS, and have only been described in one non-published case [23]. Peeters et al. suggested that the co-expression of tenascin-C (TNC) in large blood vessels might compensate for the TNX-deficiency, giving rise to a normal arterial vessel wall [37,38]. Conversely, mitral valve involvement (i.e., mitral valve prolapse, thickening and/or insufficiency) has thus far been reported in 20% of clEDS patients, highlighting the need for cardiovascular follow-up (Table 1).

#### **5. Conclusions**

Although TNX-deficiency phenotypically resembles cEDS, absent atrophic scarring, the presence of short broad feet, brachydactyly, edema of the lower extremities, acrogeria or the occurrence of hollow organ perforation should initiate targeted diagnostics for clEDS, either by measuring the TNX concentration in serum or by mutation analysis of the *TNXB* gene.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/11/843/s1, Figure S1: Steady-state analysis of collagen in the medium and the cell layer, Table S1: 101 genes analyzed by means of an NGS panel for connective tissue disorders, Table S2: Primer sequences and PCR conditions for *TNXB* sequencing, Table S3: Published *TNXB* pathogenic variants.

**Author Contributions:** M.C., M.R. (Marianne Rohrbach) and C.G. conceived the study and edited and coordinated the manuscript. The manuscript was drafted by D.R. and M.R. (Marco Ritelli), who organized data contents, reviewed the literature and completed figures and tables. D.R. and M.R. (Marianne Rohrbach) performed the clinical evaluation of the patient. M.R. (Marco Ritelli), N.Z., V.C. and C.G. provided the experimental data. All authors discussed, read and approved the manuscript.

**Funding:** This project was supported by radiz (rare disease initiative Zurich, Switzerland) to D.R. and has been supported by tshe Swiss National Science Foundation Grant No. 31003A-173183 to C.G. and M.R. (Marianne Rohrbach).

**Acknowledgments:** The authors want to thank the patient for her cooperation during the diagnostic process and the Fazzo Cusan family for its generous 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/).

## **Novel** *TNXB* **Variants in Two Italian Patients with Classical-Like Ehlers-Danlos Syndrome**

**Lucia Micale 1,**†**, \*, Vito Guarnieri 1,**†**, Bartolomeo Augello 1, Orazio Palumbo 1, Emanuele Agolini 2, Valentina Maria Sofia 2, Tommaso Mazza 3, Antonio Novelli 2, Massimo Carella <sup>1</sup> and Marco Castori <sup>1</sup>**


Received: 26 October 2019; Accepted: 21 November 2019; Published: 25 November 2019

**Abstract:** *TNXB*-related classical-like Ehlers-Danlos syndrome (*TNXB*-clEDS) is an ultrarare type of Ehlers-Danlos syndrome due to biallelic *null* variants in *TNXB*, encoding tenascin-X. Less than 30 individuals have been reported to date, mostly of Dutch origin and showing a phenotype resembling classical Ehlers-Danlos syndrome without atrophic scarring. *TNXB*-clEDS is likely underdiagnosed due to the complex structure of the *TNXB* locus, a fact that complicates diagnostic molecular testing. Here, we report two unrelated Italian women with *TNXB*-clEDS due to compound heterozygosity for *null* alleles in *TNXB*. Both presented soft and hyperextensible skin, generalized joint hypermobility and related musculoskeletal complications, and chronic constipation. In addition, individual 1 showed progressive finger contractures and shortened metatarsals, while individual 2 manifested recurrent subconjunctival hemorrhages and an event of spontaneous rupture of the brachial vein. Molecular testing found the two previously unreported c.8278C > T p.(Gln2760\*) and the c.(2358 + 1\_2359 − 1)\_(2779 + 1\_2780 − 1)del variants in Individual 1, and the novel c.1150dupG p.(Glu384Glyfs\*57) and the recurrent c.11435\_11524+30del variants in Individual 2. mRNA analysis confirmed that the c.(2358 + 1\_2359 − 1)\_(2779 + 1\_2780 − 1)del variant causes a frameshift leading to a predicted truncated protein [p.(Thr787Glyfs\*40)]. This study refines the phenotype recently delineated in association with biallelic *null* alleles in *TNXB*, and adds three novel variants to its mutational repertoire. Unusual digital anomalies seem confirmed as possibly peculiar of *TNXB*-clEDS, while vascular fragility could be more than a chance association also in this Ehlers-Danlos syndrome type.

**Keywords:** Classical-like; Ehlers-Danlos syndrome; haploinsufficiency; tenascin-X; *TNXB*

#### **1. Introduction**

Ehlers-Danlos syndromes (EDS) are a clinically variable and genetically heterogeneous group of hereditary connective tissue disorders mainly featured by abnormal skin texture and repair, tissue and vascular fragility, and joint hypermobility. The 2017 international classification identifies 13 EDS types due to deleterious variants in 19 different genes [1]. More recently, a 14th type of EDS with features overlapping classical type and due to recessive variants in the *AEBP1* gene was added to this nosology [2,3]. Among them, classical, vascular, and hypermobile EDS are the most common, while the others are rarer and their frequency in the general population remains mostly unknown.

Classical-like EDS due to biallelic variants in *TNXB*, encoding tenascin-X (*TNXB*-clEDS), is a second EDS type resembling classical EDS but distinguished from the latter by recessive pattern of inheritance and lack of atrophic scarring [1]. Tenascin-X is a large extracellular matrix–forming glycoprotein, whose deficiency was firstly involved in the etiology of EDS in 1997, by the description of a novel contiguous gene syndrome combining congenital adrenal hyperplasia (CAH) and EDS, and due to the deletion of the *CYP21A2* (OMIM 201910) and *TNXB* neighbouring genes [4]. Subsequently, it was clear that this rare type of EDS is a recessive disorder caused by homozygous or compound heterozygous *null* alleles in *TNXB* [5]. Only 24 individuals with *TNXB*-clEDS have been described to date and most of them are of Dutch origin [6]. A recent cross-sectional study on 17 individuals suggests possible phenotypic hallmarks of *TNXB*-clEDS and hypothesizes that it is underdiagnosed with EDS-like symptoms outside The Netherlands due to the complex molecular structure of the *TNXB* locus [7].

*TNXB* maps on chromosome region 6p23.1 within the human leukocyte antigen histocompatibility complex in a module characterized by highly homologous sequences between functional genes, *CYP21A2* and *TNXB*, and their corresponding pseudogenes *CYP21A1P* and *TNXA*. This genomic structure is prone to non-homologous recombinations. Misalignment events during meiosis result in pathogenic *CYP21A2*/*CYP21A1P* and *TNXA*/*TNXB* chimeric genes. To date, three major types of *TNXA*/*TNXB* chimera have been identified [4,8]. In particular, CAH-X chimera 1 (CH-1) and CAH-X chimera 2 (CH-2) have *TNXB* exons 35–44 and 40–44, respectively, replaced with *TNXA* [4,9,10]. CH-1 is characterized by a 120-bp deletion (c.11435\_11524 + 30del) due to the substitution of *TNXB* exon 35 by *TNXA* that is causative of tenascin-X haploinsufficiency in CAH-X CH-1; this region is the only well-documented discrepancy between *TNXB* and *TNXA* homologous portion. CH-2 is characterized by the variant c.12174C > G p.(Cys4058Trp) derived from the substitution of *TNXB* exon 40 by *TNXA* with a likely dominant negative effect [10]. The third chimera, termed CAH-X chimera 3 (CH-3), has *TNXB* exons 41–44 substituted by *TNXA* and is characterized by a cluster of three pseudogene variants: c.12218G > A p.(Arg4073His) in exon 41, and c.12514G > A p.(Asp4172Asn), and c.12524G > A p.(Ser4175Asn) in exon 43. This chimera has been reported in one patient, and its clinical significance is still preliminary [8].

Due to such a complex molecular architecture of the genomic region surrounding *TNXB*, Demirdas et al. [7] proposed a multistep molecular diagnostics workflow including: (i) a mixed approach of next-generation sequencing (NGS) for the non-homologous *TNXB* sequence and Sanger sequencing for the *TNXA*/*TNXB* homology region to exclude *TNXB* point variants and rare *TNXA*/*TNXB* chimera, (ii) followed by *TNXB* deletion/duplication analysis aimed to investigate the presence of the recurrent c.11435\_11524 + 30del resulting from the common *TNXA*/*TNXB* chimeric fusion and other rarer rearrangements. Following this approach, this group was able to identify 12 different *TNXB* variants associated with *TNXB*-clEDS [7].

Here, we describe the first two Italian individuals affected by *TNXB*-clEDS. Molecular testing investigated the full range of possible molecular mechanisms leading to *TNXB* null alleles and found three novel variants.

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

Patients were enrolled for this study after obtaining written informed consent for publishing pictures (individual 2) and clinical data (both individuals). This study was approved by the local ethical committee, and is in accordance with the 1984 Helsinki declaration and subsequent versions. Part of the results of molecular investigations presented in this work was obtained from the routine clinical diagnostic activities of the involved institutions.

#### *2.1. Sample Preparation and Next Generation Sequencing Analysis*

Genomic DNA was extracted from patients' and unaffected relatives' peripheral blood leucocytes by using Bio Robot EZ1 (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The DNA was quantified with Qubit spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Probands' DNA first underwent sequencing with a HaloPlex gene panel (Agilent Technologies, Santa Clara, CA, USA) designed to selectively capture known genes associated with the various forms of EDS (*ADAMTS2, AEBP1, B3GALT6, B4GALT7, CHST14, COL1A1, COL1A2, COL3A1, COL5A1, COL5A2, COL12A1, C1R, C1S, DSE, FKBP14, PLOD1, FLNA, PRDM5, SLC39A13, TNXB* and *ZNF469*) according to the current nosology. For *TNXB*, NGS sequencing is effective for exons 1 to 31 only, which correspond to the non-homologous *TNXB* sequence. Libraries were prepared using the Haloplex target enrichment kit (Agilent Technologies, Santa Clara, CA, USA) following manufacturer's instructions. Targeted fragments were then sequenced on a MiSeq sequencer (Illumina, San Diego, CA, USA) using a MiSeq Reagent kit V3 300 cycles flow cell. Reads were aligned to the GRCh37/hg19 reference genome by BWA (v.0.7.17). BAM files were sorted by SAMtools (v.1.7) and purged from duplicates using Mark Duplicates from the Picard suite (v.2.9.0). Mapped reads were locally realigned using GATK 3.8. Reads with mapping quality scores lower than 20 or with more than one-half nucleotides with quality scores less than 30 were filtered out. The GATK's Haplotype Caller tool was used to identify variants. Variants were annotated by ANNOVAR, with information about allelic frequency (1000 Genomes, dbSNP 151, GO-ESP 6500, ExAC, TOPMED, GnomAD, NCI60, COSMIC), reported or computationally estimated pathogenicity (ClinVar, HGMD, LOVD, or SIFT, Polyphen2, LRT, MutationTaster, MutationAssessor, FATHMM, PROVEAN, VEST3, MetaSVM, MetaLR, M-CAP, CADD, DANN, fathmm-MKL, Eigen, GenoCanyon), and amino acids conservation (fitCons, GERP++, phyloP100way, phyloP20way, phastCons100way vertebrate, phastCons20way mammalian, SiPhy 29way). Selected variants were interpreted according to the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMGG/AMP) [11]. Specifically, variants without clinical significance at the time of reporting (i.e., benign and likely benign) were excluded by the presence of one or more criteria for benignity. Variants which passed this preliminary selection were selected for further investigation and classified as pathogenic, likely pathogenic or uncertain significance by using the following criteria: (i) null (nonsense, frameshift) variant in a gene previously described as disease-causing by haploinsufficiency or loss-of-function; (ii) missense variant located in a critical and well-established functional domain; (iii) variant affecting canonical splicing sites (i.e., ±1 or ±2 positions); (iv) variant absent in allele frequency population databases; (v) variant reported in allele frequency population databases but with a minor allele frequency significantly lower than the known disease frequency in the general population; (vi) variant predicted as pathogenic/deleterious in ClinVar and/or LOVD; (vii) missense variant predicted as pathogenic/deleterious in most (≥75%) of the selected *in silico* predictors; (viii) variant co-segregating in two or more affected relatives; ix) the predicted pathogenic effect has been confirmed by an appropriate functional study/studies.

#### *2.2. Sanger Sequencing.*

The variants identified by NGS were confirmed by Sanger sequencing. Primer sequences are reported in Table 1. The amplified products were subsequently purified by using ExoSAP-IT PCR Product Cleanup Reagent (Thermo Fisher Scientific, Waltham, MA, USA) and sequenced by using BigDye Terminator v1.1 sequencing kit (Thermo Fisher Scientific, Waltham, MA, USA). The fragments obtained were purified using DyeEx plates (Qiagen, Hilden, Germany) and resolved on ABI Prism 3130 Genetic Analyzer (Thermo Fisher Scientific, Waltham, MA, USA). Sequences were analyzed using the Sequencer software (Gene Codes, Ann Arbor, MI, USA). The identified variant was resequenced in independent experiments.


**Table 1.** Primers used in this study.

#### *2.3. Analysis of the TNXA*/*TNXB Homology Region*

For the analysis of the *TNXA*/*TNXB* homology region, the whole genomic sequence of *TNXB,* encompassing exons 32–44 was amplified by employing a Long-PCR reaction using the primers reported in Table 1 and the protocol previously reported [7]. The PCR reaction was performed in a total 50 uL volume reaction containing 5 uL Buffer 1 (10 ×), dNTP (0.625 mM final concentration), primers (10 pmol each), 3U Expand Long Template Enzyme mix (Roche, San Francisco, CA, USA). Cycling conditions were as follows: initial denaturation at 95 ◦C, 3 min, followed by 30 cycles of 95 ◦C, 30 s, 62 ◦C, 30 s, 72 ◦C, 11 min, final elongation at 72 ◦C, 7 min. The 4987 bp PCR product was checked on ethidium bromide (EtBr) stained 1% agarose gel and then used as template for seven nested PCR amplifications with the primers listed in Table 1. Reactions were performed in a 25 uL volume reaction containing 2.5 uL Buffer (10 ×), dNTP (0,25 mM final concentration), primers (10 pmol each), 1.25 U AmpliTaq Gold DNA Polymerase (Thermo Fisher Scientific, Waltham, MA, USA). Cycling conditions were the following: initial denaturation at 95 ◦C, 10 min, followed by 30 cycles of 95 ◦C, 30 s, 60 ◦C, 30 s, 72 ◦C, 30 s, final elongation at 72 ◦C, 7 min. The overlapping Nested PCR products were then checked on EtBr stained 2% agarose gel, purified with ExoSap-IT kit (Thermo Fisher Scientific, Waltham, MA, USA) and sequenced with a ready reaction kit (BigDye Terminator v1.1 Cycle kit, Thermo Fisher Scientific, Waltham, MA, USA).

#### *2.4. Multiplex Ligation-Dependent Probe Amplification (MLPA) and Quantitative PCR (qPCR) Analysis*

MLPA was carried out using the commercial kit (SALSA MLPA KIT P155-D2 Ehlers-Danlos syndrome III & IV, MRC Holland, Amsterdam, The Netherlands). The kit includes 17 probes for *TNXB* with amplification products between 130 and 490 nucleotides. This kit also comprises two probes mapping within the region upstream of *TNXB*, located in the *ATF6B* and *BAK1* genes, and two additional probes mapping within *CYP21A2*. Complete probe sequences and the identity of the genes detected by the reference probes is available online. Hybridization, ligation, and PCR amplification were performed according to the manufacturer's instructions. DNAs from three healthy individuals were included as controls. Coffalyser. Net software (MRC Holland, Amsterdam, The Netherlands) was used for data analysis. Detected deletion was confirmed by qPCR. Primers designed for the amplification of DNA fragments by qPCR, including *TNXB* exon 4 to 8 probes, were listed in Table 1. The qPCR was performed using Power SYBR Green PCR Master Mix (Thermo Fisher Scientific, Waltham, MA, USA) on ABI 7900HT real time PCR system (Thermo Fisher Scientific, Waltham, MA, USA). Control DNA fragments were located on different chromosomes. Samples were run in triplicate using standard conditions.

#### *2.5. Chromosome Microarrays Analysis (CMA)*

CMA of the individual 1 was performed using the CytoScan™ XON array (Thermo Fisher Scientific, Waltham, MA, USA). This array contains 6.85 million empirically selected probes for whole-genome coverage including 6.5 million copy number probes and 300,000 SNP probes for LOH analysis, duo-trio assessment. The sensitivity of the platform is 95% for the detection of exon-level copy number variations (CNVs). The CytoScan™ XON assay was performed according to the manufacturer's protocol, starting with 100 ng of patient DNA. Data analysis was performed using the Chromosome Analysis Suite software version 4.2 (Thermo Fisher Scientific, Waltham, MA, USA). A CNV was validated if at least 25 contiguous probes showed an abnormal log2 ratio.

#### *2.6. Conservation of the Variant.*

Evolutionary conservation of p.Gln2760 residue of tenascin-X was investigated with protein sequence alignment generated by Clustal Omega and compared with data provided by UC Santa Cruz Genome Browser

#### *2.7. Total RNA Analysis.*

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. PCR amplification and Sanger sequencing were carried out with primers listed in Table 1.

#### *2.8. Variant Designation.*

Nucleotide variant nomenclature follows the format indicated in the Human Genome Variation Society (HGVS) recommendations. DNA variant numbering system refers to cDNA. GenBank Accession Number NM\_019105.6 was used as reference sequences for *TNXB*. Nucleotide numbering uses +1 as the A of the ATG translation initiation codon in the reference sequence, with the initiation codon as codon 1.

#### **3. Results**

#### *3.1. Case Report: Individual 1*

This is a 25-year-old woman, first child of healthy and unrelated parents. Her younger brother was described as unaffected. She was born at term, from an uneventful pregnancy and uncomplicated vaginal delivery. Birth parameters were normal. Psychomotor development and education were within normal limits. She recalled congenital contortionism (i.e., positive five-point questionnaire). At 3 years of age, the mother noted trigger fingers of multiple digits in both hands, and this was treated surgically at 16 years of age with complete resolution. She also requested ortodontic therapy for dental crowding and high-arched palate. No visual nor auditory troubles were registered. The patient suffered from constipation since infancy. At 14 years of age, she received the diagnosis of rectal prolapse. She always suffered from easy bruising and tendency to arterial hypotension. More recently, the patient requested rheumatological consultation for proneness to soft-tissue injuries and easy fatigability, although she did never complain of dislocations and fractures. According to the patient's past clinical history, the rheumatologist requested medical genetics consultation for a suspicion of hereditary connective tissue disorder. At examination, she displayed normal upper limb span/height ratio, bilaterally positive wrist

sign, bilaterally negative thumb sign, Beighton score (7/9) with genua recurvata, marked hypermobility of the fingers and toes, flatfeet, brachydactyly (due to shortened metatarsals—brachydactyly type E), of the 2nd and 3rd toes bilaterally, brachydactyly type D of the hands (i.e. shortened and broad distal phalanx of the thumbs), soft, doughy and significantly hyperextensible skin, absence of distrophic scars, palmoplantar hyperlinearity, mild palpebral ptosis and bilateral euryblepharon. Of relevance, no other close family members had brachydactyly. Heart ultrasound showed minimal insufficiency of the mitral valve. Bone densitometry showed reduced bone mass at the femoral head (T-score-1.7). Hand X-rays showed "swan neck" deformity of the interphalangeal distal joint of 1st, 2nd and 3rd left fingers.

#### *3.2. Case Report: Individual 2*

This is a unique child of unaffected and unrelated parents. Pregnancy and delivery were uneventful, and psychomotor development and education were normal. She came at our medical genetics outpatient service at 26 years of age due to a long history of polyarticular intense pain (visual analogic scale: 7/10) and (sub)luxations of the shoulders, fingers and temporomandibular joints. Additional musculoskeletal and neurological symptoms included moderate pain (visual analogic scale 4-5/10) of the spine, hands and feet, as well as recurrent myalgias, occipital headache, occasional migraine with aura, post-exertional malaise and chronic fatigue. She also reported easy bruising, an episode of spontaneous rupture of the right brachial vein, multiple episodes of ruptures of cystic lesions of the elbows with leakage of a yellowish and filamenotous substance (presumably, molluscoid pseudotumors), eye dryness, recurrent subconjunctival hemorrhages, gastroesophageal reflux with cardia incontinence, esophageal erosions and severe constipation with coprostasis. At examination, at 26 years of age, the patient showed normal anthropometry, Beighton score 7/9, marked hypermobility of the digits, temporomandibular joints and shoulders, bilateral hallux valgus, genua valga, lumbar hyperlordosis, mild thoracic scoliosis, reduced muscle tone, painful movements, soft, doughy and hyperextensible skin (Figure 1a), normal scar formation, residual elbow scars from recurrent ruptures of (likely) mulluscoid pseudotumors (not visible at the time of examination; Figure 1b), a small subcutaneous spheroid of the pretibial region on the left, multiple cutaneous hematomas, piezogenic papules (Figure 1c), palmoplantar hyperlinearity, bluish sclerae, absence of the lingual frenulum and acrocianosis. Heart ultrasound was normal. Bone densitometry showed mildly reduced bone mass at the femural neck. We were also able to carry out physical examination of individual 2's parents and both had positive Beighton score according to age and sex (i.e., generalized joint hypermobility). The father also showed mildly soft and hyperexensible skin, although both did not report any significantly related complaints.

**Figure 1.** Selected clinical features of individual 2, and schematic diagram showing the genomic structure of *TNXB* and the secondary structure of tenascin-X. (**a**) Markedly hyperextensible skin of the dorsum of the hands. (**b**) Residual scar from recurrent molluscoid pseudotumors of the elbow. (**c**) Piezogenic papules of the heel. (**d**) Coding regions are highlighted with white boxes and introns with black horizontal lines. Tenascin-X is characterized structurally from N-terminus to C-terminus by: (i) an N-terminus with a series of repeats that resemble epidermal growth factor (EGFL repeats); (ii) a stretch of fibronectin type III repeats (FBIII1-31); and (iii) a large C-terminal domain structurally related to fibrinogen (Fibrinogen C). Previously identified variants associated with *TNXB*-clEDS are represented in black. Variants identified in the individuals 1 and 2 are represented in blue and red, respectively.

#### *3.3. Molecular Findings: Individual 1*

NGS analysis performed on DNA from individual 1 identified the novel heterozygous c.8278C> T variant located in the exon 24 of *TNXB* (Figure 1d), which is predicted to incorporate a premature stop codon [p.(Gln2760 \*)]. No other candidate variants were found in the remaining genes. The c.8278C > T p.(Gln2760 \*) variant was not reported in major databases, including dbSNP, ExAC, 1000 Genomes, and GnomAD. This suggests that the variant represents a rare event and was interpreted as likely pathogenic according to the ACMGG/AMP criteria (i.e., a variant absent in population databases and predicted to generate a null allele in a gene previously known as disease-causing with this molecular mechanism). The result was confirmed by direct Sanger sequencing of proband's DNA (Figure 2a). Protein sequences alignment of the homologous regions including the Gln2760 residue of human *TNXB* was generated by using the Clustal Omega tool and showed that the affected residue was evolutionarily conserved (Figure 2b). The Gln2760 residue is located in the 19th fibronectin domain

of tenascin-X (Figure 1d) and, thus, the truncated protein presumably loses the last multiple 19–31 fibronectin domains as well as the fibrinogen C motif.

As *TNXB*-clEDS is caused by a complete lack of tenascin-X due to biallelic inactivating variants in *TNXB*, in order to detect the potential presence of a second variant in *TNXB*, we simultaneously performed the long PCR/Sanger sequencing analysis of the *TNXA*/*TNXB* homologous region and MLPA analysis. The long PCR/Sanger sequencing analysis did not reveal any variant in exons 32–44. On the contrary, MLPA analysis detected a *TNXB* intragenic deletion which includes entirely the exon 5 (Figure 2c). To narrow the proximal deletion breakpoints within the region encompassing the exon 5, qPCR analysis was employed on DNA extracted from patient's, unaffected parents, and control individuals blood. This approach detected a *TNXB* deletion which include both the whole exons 5 and 6 (Figure 2d).

To better molecularly refine the extension of the deletion, we performed a chromosome microarrays analysis using the CytoScan™ XON array. CMA confirmed an interstitial heterozygous microdeletion at chromosome 6p21.33, covered by 62 array probes and spanning 5 Kb, which encompasses the exons 5 and 6 and flanking intronic regions of *TNXB* (Figure 2e). The molecular karyotype of the patient, accordingly with the International System for Human Cytogenomic Nomenclature 2016 was arr [GRCh37] 6p21.33 (32056115\_32061375) x1.

Next, we characterized the deletion at the transcriptional level by direct DNA sequencing of in vitro amplified cDNA product generated from total RNA extracted from patients' peripheral blood leucocytes (Figure 2f). We showed that the variant c.(2358 + 1\_2359 − 1)\_(2779 + 1\_2780 − 1)del generates a frameshift with the insertion of a premature stop codon in exon 7 [p.(Thr787Glyfs\*40)] (Figure 1d). In light of its absence in population databases, the predicted generation of a null allele and the subsequent demonstration by mRNA study, this variant was interpreted as pathogenic according to the ACMGG/AMP rules. Segregation analysis in both unaffected parents was performed by Sanger sequencing and MLPA/qPCR analysis. We detected the c.8278C > T and (2358 + 1\_2359 − 1)\_(2779 + 1\_2780 − 1)del variants in the proband's father and mother, respectively (Figure 2a,d,f). Both *TNXB* c.8278C > T p.(Gln2760 \*) and (2358 + 1\_2359 − 1)\_(2779 + 1\_2780 − 1)del p.(Thr787Glyfs\*40) variants have been submitted to the LOVD (Leiden Open Variation Database, https://databases.lovd.nl/shared/ variants/0000598484, individual ID # 00266303 https://databases.lovd.nl/shared/variants/0000598485, individual ID # 00266303, respectively).

**Figure 2.** Molecular findings of individuals 1 and 2. (**a**) Electropherograms showing DNA sequencing analysis of PCR product amplified with primers targeting exon 8 of *TNXB.* Nucleotide sequences are provided. The position of the identified variant is indicated with an asterisk. (**b**) Protein sequence alignment of *TNXB* generated by Clustal Omega showed that the affected Gln 2760 residue of tenascin-X is evolutionary conserved. (**c**) Bar chart generated by Coffalyser.net- MLPA analysis software. MLPA was performed on DNA from the individual 1, her unaffected parents and controls (data not showed). A probe ratio of 1 indicates a normal DNA copy number; a probe ratio of 0.5 indicates a heterozygous deletion. MLPA analysis reveals a deletion of exon 5 of *TNXB* in individual 1. (**d**) Profiles of qPCR assay performed to map the deletion breakpoints within the region encompassing the exons 3 to 8 of *TNXB*. Relative DNA quantity of each exon was determined for the patient (red), her asymptomatic

mother and father, (green and purple, respectively), and a pool of DNA controls (CNTs, orange). (**e**) Results of chromosomal microarray analysis in the Individual 1. Intensity data (Summarized log 2 ratio value) of each probe is drawn along chromosome 6 from 32,000 to 32,080 kb (USCS Genome Browser build February 2009, hg19). The red box indicates the interstitial microdeletion (62 probes with decreased signal) identified, encompassing the exons 5 and 6 of the TNXB gene (lower panel). (**f**) Electropherograms showing cDNA Sanger sequencing of a transcript region of *TNXB* amplified with primers targeting exon 3 to 8 of Individual 1 and her mother. (**g**) Sanger sequence of a PCR product amplified with primers targeting exon 3 of *TNXB* of individual 2 and her unaffected parents. The position of the identified variant is indicated with an asterisk. (**h**) Bar chart generated by Coffalyser.ne-MLPA analysis performed on DNA from the individual 2, her unaffected father and controls (data not showed). MLPA analysis reveals the common partial deletion of exon 35 of *TNXB* in individual 2.

#### *3.4. Molecular Findings: Individual 2*

NGS platform targeted for EDS genes revealed that individual 2 carries out a single base deletion c.1150dupG located in exon 3 of *TNXB* (Figure 1d). This heterozygous variant was predicted to generate a premature stop codon at residue 441 [p.(Glu384Glyfs\*57)]. No other candidate variants were found in the remaining genes. The c.1150dupG variant was not reported in major databases. Therefore, the variant was interpreted as likely pathogenic according to the ACMGG/AMP guidelines. This result was confirmed by direct Sanger sequencing of proband's DNA. The novel variant was also detected in the proband's mother while it was absent in the father (Figure 2g). MLPA analysis (Figure 2h) detected the recurrent pseudogene-derived 120 bps deletion including the exon 35, previously described by Schalkwijk et al. as the likely result of a common *TNXA*/*TNXB* fusion gene (CAHX-CH1). This variant was inherited from the healthy carrier father.Both TNXB c.1150dupG p.(Glu384Glyfs\*57) and c.11435\_11524 + 30del variants have been submitted to the LOVD (https://databases.lovd.nl/shared/ variants/0000598486, individual ID # 00266304¸ https://databases.lovd.nl/shared/variants/0000598487, individual ID #00266304, respectively).

#### **4. Discussion**

In this work, we described the first two Italian patients with *TNXB*-clEDS, confirming a wider distribution of this rare EDS type in Europe, and the efficacy and reproducibility of the diagnostic approach published by Demirdas et al. [7]. We also identified three novel disease-alleles in *TNXB*, which expand the known mutational spectrum of *TNXB* associated with clEDS (Figure 1d).

These two adults manifest the previously defined phenotypic spectrum of *TNXB*-clEDS. Scarring was apparently normal in our patients, which is in line with the lack of atrophic/dystrophic scarring as a distinguishing feature, together with recessive inheritance, of *TNXB*-clEDS from classical EDS. Intriguingly, individual 2 reported a history of recurrent ruptures of elbow cystic lesions resembling molluscoid pseudotumors, which are additional cutaneous features considered highly suggestive for classical EDS. This expands the cutaneous similarities between *TNXB*-clEDS and classical EDS; a fact that complicates the differential diagnosis of these disorders on clinical groups and reinforces the opportunity to consider *TNXB* molecular testing in all individuals with a clinical diagnosis of classical EDS resulted negative to *COL5A1*, *COL5A2,* and *COL1A1* (recurrent variants) analysis. Individual 2 also testifies for a possible vascular involvement in *TNXB*-clEDS. In fact, this patient reported recurrent subconjunctival hemorrhages, a feature previously annotated in multiples subjects by Demirdas et al. [7], as well as spontaneous rupture of the brachial vein. The latter is an apparently novel feature of *TNXB*-clEDS and could indicate, if confirmed by other observations, a more severe vascular involvement in this condition.

Demirdas et al. [7] pointed out a peculiar appendicular phenotype of *TNXB*-clEDS featured by foot brachydactyly and small joint (apparently acquired) contractures. Our individual 1 supports this hypothesis, as she showed shortened metatarsals and a history of multiple acquired finger contractures

with residual swan neck deformities of the left fingers. Furthermore, constipation and evacuation troubles represented major complaints in both individuals. While these features are not rare within the EDS community of syndromes and are highly represented in adults with hypermobile EDS and hypermobility spectrum disorders [12], this observation in *TNXB*-clEDS confirms the opportunity to better investigate the long-range manifestations of these disorders in order to improve quality of life of individuals with EDS.

To date, a total of 15 different *TNXB* deleterious variants, including the three novel reported in this paper were described. These variants are frameshift (7/15), stop codon (4/15), or splicing (2/15) and lead to the insertion of a premature stop codon with a presumed loss of expression of the protein. Two out of these 15 variants (2/15) are missense which have detrimental effects on the proper protein folding and stability [10,13]. Among the 15 variants, 11 are identified in single patients/families. A 2 bp deletion (c.3290\_3291del), a 30Kb deletion generating a *TNXB*/*TNXA* fusion gene, and a pseudogene-derived missense variant (c.12174C > G) were found in more than one patient (Figure 1d). Nevertheless, the current nosology of EDS and related disorders clearly states that only "*null* alleles" in *TNXB* can be considered causative of *TNXB*-clEDS [1,6]. Therefore, missense *TNXB* variants should be considered supportive of the diagnosis in a clinical setting only if they appear convincing for haploinsufficiency.

In this study, individual 1 carries two novel *TNXB* variants, c.8278C>T located in the exon 24 and (2358 + 1\_2359 − 1)\_(2779 + 1\_2780 − 1)del which results in the non in frame deletion of whole exons 5 and 6. Both variants are predicted to generate a premature stop codon. Individual 2 is a compound heterozygote for c.1150dupG and c.11435\_11524 + 30del variants. The c.1150dupG is a novel variant located in exon 3 of *TNXB* and is predicted to create a premature stop codon. The c.11435\_11524 + 30del variant, which abolishes part of exon 35 and intron 35 of *TNXB,* has been previously described and *TNXA*/*TNXB* chimeric recombination type 1 [5,7,10]. This deleted region represents the only large *TNXB*-specific sequence in the *TNXA*-homolog region of *TNXB*. The *TNXA*-derived variation is a molecular event which often takes place between a functional gene and a pseudogene. Although this *TNXA*/*TNXB* fusion gene has been previously characterized, its molecular effect is not yet known. However, tenascin-X serum measurement in affected individuals by previous studies indicate that this variant likely results in a *null* allele [5,9]. In this work, we were not able to carry out a serum dosage of tenascin-X in our patients. Nevertheless, we are confident that the molecular features of the identified variants are convincing for the generation of a not functional allele.

Due to the complex nature of the genomic region spanning around *TNXB*, the underdiagnosis of *TNXB*-clEDS in the routine diagnostic activities of most laboratories is a likely scenario. In fact, molecular testing of *TNXB* is challenging due to the presence of the pseudogene *TNXA,* which is more than 97% identical to *TNXB* at its 3 end (exons 32–44). With the only exception of exon 35, which partially shows a *TNXB* specific sequence (see above), exon and intron sequences in this region are (nearly) identical in both *TNXB* and *TNXA*. Our experience confirms the need of a multistep and multi-technique approach (comprising NGS for the non-homologous region, Sanger sequencing with a long-PCR and nested-PCR system for the *TNXA*/*TNXB* homologous segment, and quantitative analysis for intragenic and intergenic rearrangement) for an efficient analysis of the entire *TNXB* coding region, with only slight modifications from the methodology proposed by Demirdas et al. [7] Tenascin-X serum concentration measurement in patients with suspected *TNXB*-clEDS is an alternative in the absence of effective molecular diagnostic facilities.

In summary, we reported two additional individuals with *TNXB*-clEDS. Our findings support the previously defined phenotype, which shows similarities with classical EDS but also include some possible distinguishing features and potentially underreported, clinically relevant manifestations. We also expanded the mutational spectrum of *TNXB* and highlighted the need of a high level of specialty for an efficacious *TNXB* molecular screening in a clinical setting.

**Author Contributions:** M.C. (Marco Castori), L.M., and V.G. designed the study and wrote the manuscript. L.M., V.G., E.A., and A.N. performed NGS analysis and studies on cDNA. B.A., E.A., V.M.S., and A.N. performed MLPA and qPCR analysis. O.P. and M.C. (Massimo Carella) performed CMA analysis. M.C. (Marco Castori), V.G., and L.M. interpreted molecular data. T.M. carried out the bioinformatics analysis. M.C. (Marco Castori) provided

clinical evaluation of the patients. All authors contributed to the writing and reviewing the manuscript and approved its final version.

**Funding:** This work was supported by the Ricerca Corrente 2018–2019 Program from the Italian Ministry of Health.

**Acknowledgments:** The authors thank the family for their kind availability in sharing the findings within the scientific community.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

#### **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/).

### **Absence of Collagen Flowers on Electron Microscopy and Identification of (Likely) Pathogenic** *COL5A1* **Variants in Two Patients**

**Chloe Angwin 1, Angela F. Brady 1, Marina Colombi 2, David J. P. Ferguson 3,4, Rebecca Pollitt 5, F. Michael Pope 6, Marco Ritelli 2, Sofie Symoens 7, Neeti Ghali <sup>1</sup> and Fleur S. van Dijk 1,\***


Received: 25 August 2019; Accepted: 20 September 2019; Published: 27 September 2019

**Abstract:** Two probands are reported with pathogenic and likely pathogenic *COL5A1* variants (frameshift and splice site) in whom no collagen flowers have been identified with transmission electron microscopy (TEM). One proband fulfils the clinical criteria for classical Ehlers-Danlos syndrome (cEDS) while the other does not and presents with a vascular complication. This case report highlights the significant intrafamilial variability within the cEDS phenotype and demonstrates that patients with pathogenic *COL5A1* variants can have an absence of collagen flowers on TEM skin biopsy analysis. This has not been previously reported in the literature and is important when evaluating the significance of a TEM result in patients with clinically suspected cEDS and underscores the relevance of molecular analysis.

**Keywords:** classical Ehlers-Danlos Syndrome; electron microscopy; collagen flowers; *COL5A1*

#### **1. Introduction**

The Ehlers-Danlos syndromes (EDS) consist of 13 subtypes with overlapping features including joint hypermobility, skin, and vascular fragility and generalised connective tissue friability [1,2]. Current major criteria for classical EDS (cEDS) are (1) skin hyperextensibility and atrophic scars and (2) joint hypermobility. Minor criteria are easy bruising, soft doughy skin, skin fragility, molluscoid pseudotumors, subcutaneous spheroids, hernia(s), epicanthal folds, complications of joint hypermobility, and an affected first degree relative. The minimal criteria for a diagnosis of cEDS are major criterion 1 plus either major criterion 2 or 3 of the 9 minor criteria [3].

In patients who satisfy the main criteria of cEDS according to the Villefranche criteria [4], the variant detection rate in either *COL5A1* or *COL5A2* is over 90% [5,6]. However, intrafamilial variability in classical EDS has been reported [7]. There are 194 reported unique variants reported in the *COL5A1* gene [8]. These genes encode collagen type V, a fibrillar heterotrimer ([α1(V)]2 α2(V)) that is present in a wide variety of tissues but is particularly prevalent in bone, skin and tendon [9]. Collagen type V accounts for approximately 5% of total body collagen and has a role in maintaining the deposition and structure of other more abundant collagens, particularly collagen type I [9]. Rarely, a diagnosis of cEDS is due to dominant variants in *COL1A1* or *COL1A2* [3,10,11].

For many years, skin biopsies for electron microscopy (EM) have been recommended as a first line of investigation to confirm or exclude a diagnosis of cEDS [3,12]. This was due to the occurrence of collagen flowers visible by EM. Collagen flowers in individuals with classical EDS were described by Vogel et al. who reported collagen fibrils with an abnormally large diameter and a highly irregular and lobulated contour interspersed with normal appearing fibrils with a mean diameter larger than that of collagen fibrils in normal skin [13]. A longitudinal section showed that the large atypical fibrils were seen to be poorly integrated filamentous aggregates. Variation in frequency of very large highly irregular fibrils differed per patient but in general constituted approximately 5% [13]. These very large highly-irregular fibrils are often described as longitudinally splayed and loosely packed fibrils, which in cross section produce the collagen flower pattern. Although it is known that collagen flowers can be found in other collagen disorders including osteogenesis imperfecta and Ullrich congenital muscular dystrophy, typical collagen flowers were thought to be invariably present in people with cEDS [3,14,15]. Given the high detection rates of pathogenic variants in cEDS, current recommendations are that electron microscopy based on a skin biopsy should no longer the first line of investigation but could be used to clarify inconclusive molecular results, or to guide further testing if initial molecular testing is negative [3]. Interestingly, it was mentioned that "the absence of typical collagen flowers would go against the diagnosis, as there are no known reports of patients with type V collagen abnormalities without collagen flowers on EM" [3]. Here, we report for the first time two patients, one fulfilling the clinical diagnosis of classical EDS and one not, with (likely) pathogenic variants in *COL5A1* and absence of collagen flowers on EM.

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

#### *2.1. Skin Biopsy*

Skin biopsies in probands 1 and 2 were taken in the form of a punch biopsy from the medial surface of the arm. P1 was 44 years and P2 was 39 years when the skin biopsy was performed. The sample for transmission electron microscopy (TEM) (3 mm) was preserved in 4% glutaraldehyde in 0.1 M phosphate buffer. This publication does not constitute research and does not require formal Research Ethics approval or Research and Development Approval as stipulated by the UK Policy Framework for Health and Social Care Research and the Health Research authority decision tool.

#### *2.2. DNA Analysis*

Sanger sequencing of the *COL5A1* and *COL5A2* genes and MLPA of the *COL5A1* gene was performed in proband 1. Sanger sequencing and MLPA of the *COL3A1* gene was performed in proband 2. Proband 2 was included in a large research project aimed at identifying new pathogenic variants with next-generation sequencing in patients with an EDS phenotype [16]. The sequencing data of patient 2 have been deposited to the EDS and OI databases: https://www.le.ac.uk/genetics/collagen.

#### **3. Clinical Report**

#### *3.1. Family 1*

Proband 1 is a 47-year-old woman, born at 36 weeks gestation, with a childhood history of clumsiness, tissue fragility, and joint hypermobility (Figure 1A–C). She would bruise easily and recalls some episodes of significant swelling after mild injury, occasionally requiring drainage when this involved the knees. There was no history of dislocation or fracture. As an adult she reports myalgia and arthralgia in the ankles, hips, wrists, and shoulders. Other medical history includes endometrial polyps removed without complication at age 32, varicose veins treated surgically at age 41, hiatus hernia, diverticular disease, a urethral cyst, and hypotension associated with fainting episodes. Neurocardiology investigations found a high tendency towards POTS (postural orthostatic tachycardia syndrome) with no evidence of autonomic failure. Cardiac investigations have shown right bundle branch block and trivial aortic regurgitation, tricuspid regurgitation, and mitral regurgitation.

She fulfilled the criteria for cEDS on physical examination. There were no craniofacial dysmorphic features except for epicanthic folds. She had soft, doughy, and hyperextensible skin—particularly at elbows, neck, and knees—with redundancy at the knees and Achilles tendons. She had scarring over the forehead, wrist, and lower legs. Lower leg scars were widened and atrophic with erythema and hemosiderin deposition. There were thread veins and varicose veins in the lower limbs. Her feet had bilateral hallux valgus deformities with piezogenic papules. Hands had bilateral hitchhikers' thumb and increased palmar markings. She had generalized joint hypermobility (gJHM) with a (Beighton score 5/9).

Family history: The proband's parents, four older brothers and their children are healthy with no reported features of cEDS. One brother was tested for the *COL5A1* variant as well. He did not meet the criteria for cEDS and was found not to have the *COL5A1* variant.

**Figure 1.** Pedigrees and clinical photographs of probands 1 and 2. (**A**) Family pedigree of proband 1. E1 denotes clinical diagnosis of cEDS, E2 denotes presence of (likely) pathogenic *COL5A1* variant, \* denotes documented evaluation. Shading denotes having a clinical diagnosis of cEDS and presence of (likely) pathogenic *COL5A1* variant. (**B**) Knee area of proband 1 demonstrating redundancy of skin, atrophic and widened scarring and varicose veins. (**C**) Hyperextensible skin at the elbow in proband 1. (**D**) Family pedigree of proband 2. (**E**) Knees area of proband 2 show no atrophic scarring or haemosiderin deposition but minor redundancy of skin. (**F**) Widened, atrophic scar at the elbow.

#### *3.2. Family 2*

Proband 2 is a woman who was diagnosed with an acute, spontaneous, left carotid artery dissection presenting with left sided headache, and Horner's syndrome at 37 years old (Figure 2D–F). This was successfully treated conservatively. She had no past medical history and was a non-smoker with normal cholesterol levels. Her past surgical history included bilateral bunion surgery at age 13 and three Caesarean sections, the first for fetal distress and the others as routine following without complication.

On physical examination she did not have craniofacial dysmorphic features [17]. Her skin was mildly hyperextensible over face, neck, and elbows with delayed recoil and slight redundancy of the skin around the knees. The skin was not thin and no bruising was visible. A widened, atrophic scar at a site of trauma over the elbow and thin, well healed post-operative scars from bunion surgery were observed. There was bilateral hallux valgus She did not have gJHM. (Beighton score 3/9) and did not fulfil the current clinical criteria for cEDS. She had a normal echocardiogram.

Family history: The mother of the proband has the *COL5A1* variant and had a healthy childhood but developed hypertension later in life. Findings on clinical examination at the age of 81 were that of skin hyperextensibility with significantly delayed recoiling probably due to her advanced age, abdominal striae, a mild spinal scoliosis, and bilateral hallux valgus. The father of the proband had a history of joint hypermobility and hallux valgus and passed away at the age of 81. He had a healthy childhood and later in life developed hypertension, bowel cancer, skin cancer, and cardiovascular disease, requiring a double coronary artery bypass graft post myocardial infarction.

The proband has two siblings, a sister and a brother. There is no clinical information available regarding her brother. The proband's sister has a cardiovascular history with bicuspid aortic valve requiring replacement, and aortic root dilatation. She does not have the *COL5A1* variant found in her mother and sister.

The proband has three sons, all born by caesarean section. The oldest son, 9 years old, experienced foetal distress and oxygen deprivation in the womb, requiring emergency caesarean section and developed cerebral palsy with significant spasticity. On examination, there was minimal bruising, atrophic scarring over the forehead and occiput, mild hyperextensibility of the skin, and a Beighton score of 2/9 and therefore did not meet the current clinical criteria for cEDS. He had a normal echocardiogram and was found to have the *COL5A1* variant. The middle son, 7 years old, had had a healthy childhood. On examination he had hyperextensible skin, atrophic scarring on the forehead and lower legs, bruising over the shins and a Beighton score of 7/9 and therefore met the current clinical criteria for cEDS. He had a normal echocardiogram and he was found to have the *COL5A1* variant. The youngest son, 4 years old, had had a healthy childhood and did not have any clinical features of cEDS. He therefore did not meet the current clinical criteria for cEDS and was found not to have the *COL5A1* variant.

#### **4. Results**

#### *4.1. Skin Biopsy TEM Results*

Proband 1: TEM was concluded to be relatively normal but showed some minor deformation of the outline of certain collagen fibrils (Figure 2a). No large collagen flowers were observed. Elastic fibres and fibroblasts appeared normal.

Proband 2: TEM showed collagen clumps within the reticular dermis. Within these, the collagen fibrils themselves were of relatively even diameter and had a symmetrical circular outline (Figure 2b), although longitudinally sectioned fibres showed occasional kinking. No abnormal collagen flowers were seen within the biopsy (Figure 2b). The other structures including elastic fibres and fibroblasts were normal in appearance. The conclusion was one of a normal skin biopsy. After the molecular result, re-examination identified rare, slightly deformed fibrils (Figure 2b) with slight variation in collagen fibril diameter but no further abnormalities and no typical collagen flowers as described previously (Figure 2c). It should be noted that collagen flowers are concentrated within the reticular dermis and rarely observed in the papillary dermis. Therefore, care is required to ensure the correct area of the skin biopsy was examined.

**Figure 2.** Electron micrographs of the reticular dermis of skin biopsy from proband 1 (**a**), proband 2 (**b**) and a case of classical EDS (**c**). Bars represent 100nm. (**a**) Cross section of collagen fibres from proband 1 showing a number of fibres presenting an irregular outline (arrowheads). Insert: detail of an irregular fibre. (**b**) Cross section through a clump if collagen fibres of proband 2 showing a relatively normal circular outline although very rare slightly irregular fibres were observed (insert). (**c**) Cross section of collagen fibres from a case of classical EDS showing a number of irregular fibres and large collagen flowers (arrowheads).

#### *4.2. DNA Analysis*

Proband 1: DNA analysis identified the heterozygous *COL5A1* variant c.4414del, p.(Leu1472Serfs\*16). This variant results in a pathogenic frameshift in exon 57 of *COL5A1* (NM\_000093.4; NP\_000084.3), predicted to result in a premature termination codon, nonsense-mediated RNA decay, and consequent type V collagen haploinsufficiency.

Proband 2: DNA analysis of *COL3A1* in which variants cause vascular EDS, did not identify pathogenic variants. The patient was referred for further sequencing as part of a large research project looking at Mendelian inheritance [16] and the likely pathogenic variant c.4068G>A; p.(Ala1356=) in the *COL5A1* gene was identified. This silent variant has been previously published is predicted to result in abnormal splicing of exon 51 of the *COL5A1* gene [7]. Splice site prediction software (Alamut visual software) indicated that the variant would destroy the splice donor site leading to disruption of normal splicing with skipping of exon 51, leading to the production of a shortened protein [7]. The same variant was identified in the mother of the proband as well as the proband's two older sons, associating with the clinical diagnosis of cEDS. The variant was not identified in the youngest son (Figure 1B).

#### **5. Discussion**

We report two patients with a (likely) pathogenic *COL5A1* variant who did not have collagen flowers on electron microscopy although minor deformation of the outline of certain collagen fibrils were observed in proband 1, which are similar to the initial stage of collagen flower formation (cf Figure 2a,c). Large collagen flowers on electron microscopy have previously been used to confirm or exclude a clinical diagnosis of cEDS. One other report describes a 12-year-old boy with marked clinical features of cEDS, without collagen flowers on electron microscopy. Interestingly, investigation showed disorganisation, variation of fibril diameter, and irregular fibril outlines. However, this patient did not have molecular investigations, and the underlying molecular basis is unknown [18].

#### *5.1. Phenotype*

Proband 1 clearly fulfilled the criteria for cEDS. Proband 2 did not fulfil the clinical criteria for cEDS. Moreover, she presented with a vascular complication (carotid artery dissection) which is not a common feature in cEDS [19]. In a recent systemic review, 12/110 (11%) patients with *COL5A1*/*2* variants had vascular complications of which six had arterial dissection, three had arterial aneurysms, and one had intracerebral haemorrhage. It is not certain whether the vascular complication in this proband is linked to the *COL5A1* variant. It has been hypothesized that glycine substitutions near or at the C-terminal end of collagen type V, may predispose to vascular events [20], but at this point there is not enough evidence to support this hypothesis.

There is clear intrafamilial variability across the generations of the family of proband 2, as the mother of the proband was 81 years old with skin hyperextensibility as the only major feature whereas one grandson with the same variant had hyperextensible skin and atrophic scarring, clearly fulfilling the clinical criteria for cEDS. Wide variability between family members with the identical *COL5A1* variant has been previously reported [6,17].

#### *5.2. Transmission Electron Microscopy*

In proband 1, who fulfilled the clinical criteria for cEDS, minor deformation of the outline of certain collagen fibrils were observed but no collagen flowers were present. The appearances were similar to early changes seen in fibrils in cEDS (Figure 2a,c). These changes were sufficient for the abnormality to be highlighted in the EM report but not significant enough to conclude that it concerned an abnormal biopsy. In proband 2, no collagen flowers or other abnormalities were observed. Unfortunately, the family members of proband 2, who had pathogenic *COL5A1* variant (mother and sons of proband), did not consent for a skin biopsy to be performed and as such no TEM results are available. While the presence or absence of collagen flowers is reported, it should also be noted that incidence and size/complexity of the collagen flowers can vary markedly between patients but the significance of this to the clinical features and the underlying genetic cause is unknown (Ferguson unpublished observations).

#### *5.3. COL5A1 Variants*

The variant identified in proband 1, who fulfils clinical criteria for cEDS, is a pathogenic frameshift variant in exon 57 predicted to result in a premature termination codon leading to haploinsufficiency. The likely pathogenic variant in *COL5A1* c.4068G>A identified in proband 2, who does not fulfil clinical criteria for cEDS, is a splice site variant and expected to result in skipping of exon 51 which is in-frame and as such would lead to a shortened protein and exert a dominant-negative effect. One variant was described by Symoens et al., *COL5A1* c.4068G>T which demonstrated skipping of exon 51 on mRNA analysis; p.(Gly1339\_Ala1356del). The proband had skin hyperextensibility, atrophic scarring, joint hypermobility, and no history of vascular events. TEM was not performed [5]. Colombi et al, identified an identical variant. The proband in this case did not fulfil the clinical criteria for cEDS, and reported joint instability, gastrointestinal symptoms, and soft skin with a few small atrophic scars over the knees. She had historical generalised joint hypermobility. Predictive software (Alamut visual software) projected a similar functional outcome as the previous *COL5A1* c.4068G>T variant [7], both reduce the donor site strength to similar extents so a similar protein effect would be expected. According to the ACMG guidelines [21] this variant is classified as likely pathogenic fulfilling *PP3*: Multiple lines of computational evidence support a deleterious effect on the gene or gene product (conservation, evolutionary, splicing impact, etc.); *PM2*: Absent from controls (or at extremely low frequency if recessive) in Exome Sequencing Project, 1000 Genomes Project, or Exome Aggregation Consortium and *PS1*: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change.

Hypotheses for this incomplete cEDS phenotype included protective lifestyle factors, potential for this variant to result in only partial activation of abnormal splicing of exon 51 and the existence of protective variants in other genes, which may counteract the loss of collagen V function during the deposition of structural collagens [7].

Of the 194 unique variants reported in the *COL5A1* gene, 31 have been reported as splice site variants [8]. Most pathogenic *COL5A1* variants (including splice site variants that introduce a premature stop codon) lead to haploinsufficiency for *COL5A1* mRNA. This is expected to be the case for the variant reported in P1. Structural variants exerting a dominant-negative effect are a minority and most commonly involve splice site variants resulting in exon skipping which is the case for the variant reported in P2 and variants that result in the substitution for glycine in the triple-helical region [22]. Collagen type V is thought to perform a regulatory function in collagen fibrillogenesis. It has been hypothesized that the final common pathway for all *COL5A1* variants is reduced availability of collagen type V, and that clinical phenotypes result from disrupted fibrillogenesis [5,8,23,24].

#### **6. Conclusions**

In conclusion, we present two probands with (likely) pathogenic *COL5A1* variants (frameshift and splice site) in whom no collagen flowers were identified, although minor deformation of the outline of certain collagen fibrils was observed in proband 1. Proband 1 fulfils the clinical criteria for cEDS but proband 2 does not and presents with a vascular complication. The mother and two sons of proband 2 also have the *COL5A1* variant, one of whom fulfils clinical criteria of cEDS. This case report highlights the significant intrafamilial variability within the cEDS phenotype. We demonstrate that patients with (likely) pathogenic *COL5A1* variants can have an absence of collagen flowers on biopsy. It is currently unclear whether the absence of collagen flowers can be linked to the (severity of) clinical features and/or the specific genetic cause. It is also uncertain whether the vascular complication in P2 is caused by the *COL5A1* variant. Absence of collagen flowers in patients with (likely) pathogenic *COL5A1* variants has not been previously reported in the literature but is important when evaluating the significance of a TEM result in patients with suspected cEDS and underscores the relevance of molecular analysis.

**Author Contributions:** Conceptualization, F.S.v.D., N.G.; Methodology, D.J.P.F.; Formal Analysis, C.A., D.J.P.F., R.P., S.S., F.S.v.D.; Investigation, D.J.P.F., F.M.P., N.G., and F.S.v.D.; Writing—Original Draft Preparation, C.A. and F.S.v.D.; Writing—review and editing, A.F.B., M.C., D.J.P.F., F.P., M.R., and N.G.; Visualization, C.A., D.J.P.F., and F.S.v.D.; Supervision, F.S.v.D.

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

**Acknowledgments:** We would like to thank the families for their kind cooperation.

**Conflicts of Interest:** All authors state that there is 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/).

### *Article* **Arterial Elasticity in Ehlers-Danlos Syndromes**

#### **Amanda J. Miller 1,\*, Jane R. Schubart 2, Timothy Sheehan 3, Rebecca Bascom <sup>4</sup> and Clair A. Francomano <sup>5</sup>**


Received: 13 November 2019; Accepted: 2 January 2020; Published: 4 January 2020

**Abstract:** Ehlers-Danlos Syndromes (EDS) are a group of heritable disorders of connective tissue (HDCT) characterized by joint hypermobility, skin hyperextensibility, and tissue fragility. Orthostatic intolerance (OI) is highly prevalent in EDS however mechanisms linking OI to EDS remain poorly understood. We hypothesize that impaired blood pressure (BP) and heart rate control is associated with lower arterial stiffness in people with EDS. Orthostatic vital signs and arterial stiffness were assessed in a cohort of 60 people with EDS (49 female, 36 ± 16 years). Arterial elasticity was assessed by central and peripheral pulse wave velocity (PWV). Central PWV was lower in people with EDS compared to reference values in healthy subjects. In participants with EDS, central PWV was correlated to supine systolic BP (r = 0.387, *p* = 0.002), supine diastolic BP (r = 0.400, *p* = 0.002), and seated systolic BP (r = 0.399, *p* = 0.002). There were no significant correlations between PWV and changes in BP or heart rate with standing (*p* > 0.05). Between EDS types, there were no differences in supine hemodynamics or PWV measures (*p* > 0.05). These data demonstrate that increased arterial elasticity is associated with lower BP in people with EDS which may contribute to orthostatic symptoms and potentially provides a quantitative clinical measure for future genotype-phenotype investigations.

**Keywords:** Ehlers-Danlos syndromes; pulse wave velocity; blood pressure; orthostatic intolerance

#### **1. Introduction**

Ehlers-Danlos syndromes (EDS) are a collection of heritable disorders of connective tissue characterized by joint hypermobility, mild skin hyperextensibility, and tissue fragility [1]. Common symptoms of EDS include joint instability, chronic pain, gastrointestinal issues, and sleep disturbances [2]. Many people with EDS have persistent symptoms of orthostatic intolerance (OI) including lightheadedness, fatigue, nausea, and palpitations [3]. Additionally, the prevalence of EDS is higher in patients with orthostatic intolerance compared to the general population [4]. The association between EDS and autonomic cardiovascular dysfunction is most prevalent in people with hEDS [5–7], but there is also evidence of orthostatic intolerance in classical EDS [8]. The high prevalence of OI in EDS demonstrates a need to understand cardiovascular pathophysiology in all EDS types, as the pathophysiology explaining the high rate of OI in EDS is unknown. The leading theory connecting the two disorders is that generalized connective tissue laxity in EDS increases vascular compliance, leading to insufficient vasoconstriction and venous insufficiency when upright resulting in symptoms of OI [3]. Despite its wide acceptance, there is only data to support this theory in small samples of people with vascular EDS and there is no published evidence to support this theory across other types of EDS.

Pulse wave velocity (PWV) has emerged as the gold standard method for measuring stiffness of the arteries because of its reliability and reproducibility [9]. PWV is a non-invasive technique that involves placing pressure transducers on the skin that can sense the velocity of blood traveling in the arteries, which is a function of the stiffness or elasticity of the arteries. Central PWV, the most widely used and accepted measurement for PWV, measures the stiffness or elasticity of the central cardiovascular system from the carotid to femoral arteries. Using this technique, increased PWV (implying increased arterial stiffness) has been shown to predict future hypertension, coronary heart disease, stroke, adverse cardiovascular events, and mortality [10].

While PWV is well accepted as a measure of arterial stiffness, it has been used far less often to measure arterial elasticity, which is the mathematical inverse of stiffness. More distensible arteries will stretch more as pulse waves travel, resulting in lower (slower) pulse wave velocity. Few studies have sought to identify people, including those with EDS, with suspected increased arterial elasticity and hence decreased PWV. One study evaluated PWV in nine people with comorbid hypermobile EDS and postural tachycardia syndrome (POTS) and found PWV measurements were not different compared to healthy controls [11]. Two studies examined PWV in people with vascular EDS. One study found decreased PWV in about 20% of genetically related people with vascular EDS [12]. The other study found that PWV measurements in people with vascular EDS were similar to those of healthy volunteers [13].

Therefore, the current study is the first assessment of PWV measurements in a large heterogeneous sample of people with EDS. We hypothesized that the collagen changes in EDS would confer an increased distensibility of the vasculature in all EDS types, and that this would contribute to orthostatic intolerance. In this study, we investigated central and peripheral arterial stiffness in people with EDS using the non-invasive measurement of pulse-wave velocity (PWV). We hypothesize that impaired blood pressure (BP) and heart rate control is associated with increased arterial elasticity in people with EDS.

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

The National Institute on Aging (NIA) study Clinical and Molecular Manifestations of HDCT was designed to investigate the natural history of the most common HDCT. Emphasis was placed on the cardiovascular, musculoskeletal, and neurological complications of HDCT and the natural history of these complications. The original study protocol was designed to collect clinical and family history data, and to use this information to clarify the clinical distinctions between diagnoses. Consenting participants were initially classified based on diagnostic criteria in place at the time of their clinical visit at the NIA (2001–2013). Subjects contributing only biological samples were diagnosed either through a limited onsite evaluation or through review of submitted medical records. The HDCT NIA Dataset v. 2016 consented cohort includes 1009 participants with an average age of 39 ± 18 years (range 2–95, median 40). One hundred and ninety-four participants were 18 years or younger.

The NIA study Clinical and Molecular Manifestations of HDCT began by assembling consented cohorts with a wide range of heritable HDCT, under an umbrella protocol (Protocol 2003-086, later changed to 03-AG-N330). After the study was closed to enrollment, the Institutional Review Board approved the reorganization and migration of the data into a relational database repository and approved re-contacting participants to determine if they would be interested in participating in future research. The HDCT cohort data are provided in SAS datasets, PDF, Excel, MRI DICOM file formats and are now under the umbrella of protocol 11-AG-N079, Sample and Data Repository Protocol for NIA Studies. Participants were recruited from the pool of patients previously seen by the principal investigators and from patient support groups nationally. An authorized guardian provided consent for minor participants, with age-appropriate assent by the minor. In 2016, a signed Data Transfer Agreement between NIA and Penn State University resulted in transfer of a copy of the

HDCT NIA Dataset v.2016 data repository to the Penn State University Clinical Translational Science Institute (PSU-CTSI). Datasets were accompanied by copies of original CRFs and SAS dataset codebook descriptions [14].

Participants were stratified by EDS type including: classical, hypermobile, vascular, or other or unclassified according to the Villefranche nosology [15] as previously described for this cohort [14]. Briefly, classical EDS was determined by joint laxity and skin that is extremely hyperextensible, fragile, bruises easily, and has thin atrophic scars. Hypermobile EDS was classified by history of dislocations, generalized joint laxity, and velvety texture of skin with an absence of extreme skin extensibility and profoundly abnormal scars. Vascular EDS was determined by genetic testing for variation in the COL3A1, the gene encoding type III collagen. The other and unclassified EDS category included patients with the rarer types of Ehlers–Danlos syndromes. A molecular diagnosis was used for the arthrochalasia and kyphoscoliotic types. Some patients had features overlapping with two or more types of EDS, and classification proved to be difficult in those cases, and such patients were diagnosed as "EDS, unclassified". If there was the clinical impression of EDS but they did not meet the diagnostic criteria for any of the known types, we assigned a diagnosis of "EDS, unclassified".

The analytic cohort for the present study was a subset of the EDS cohort from this NIA study of HDCT consisting of 60 participants who had both orthostatic BP recordings and PWV measurements. BP and heart rate were measured by brachial artery oscillometry in triplicate following 5 min in the supine, seated, and standing postures. Central arterial stiffness was measured by carotid to femoral PWV and peripheral stiffness by carotid to radial PWV.

#### *2.1. PWV Measurements*

The methods used to assess PWV in this study were the same methods used in the Baltimore Longitudinal Study of Aging [16]. In short, PWV data were collected using a SphygmoCor device (AtCor Medical) that utilizes an EKG and high-fidelity tonometer to acquire waveforms from carotid, femoral, and radial pulses. The software determines the velocity of the pulse wave, i.e., estimated time that it takes the pulse wave to travel between pulse sites divided by the distance between sites. Central PWV is calculated by measuring pulse waves at the carotid and femoral arteries, representing the stiffness of the central vascular tree. Peripheral PWV is calculated by measuring pulse waves at the carotid and radial arteries indicating blood flow to peripheral vascular beds. Reference values of pulse wave velocity in healthy humans were collected using similar methods (pulse wave tonometry divided by distance between sites) [16].

#### *2.2. Orthostatic Vital Sign Measurements*

Orthostatic vital signs were measured by a brachial artery BP cuff on both arms. BP was measured supine then seated then during standing. Study participants stayed in each posture (supine, sitting, and standing) for 5 min prior to BP recordings. BP was measured in triplicate in each position with one minute between recordings. If BP varied by 15 mmHg or heart rate by 10 beats/minute in one position, a fourth recording was measured. All BP and heart rate measurements on the left arm were averaged for each participant in each posture.

#### *2.3. Data Analysis*

Descriptive statistics include demographic data and EDS type. Comparison of characteristics among types was performed using ANOVA with post-hoc Tukey-Kramer tests when justified. Pearson's correlations were run between BP, heart rate, and PWV measurements for the entire cohort. We performed a stratified analysis of central PWV measurements by age in the EDS participants of all types and compared those values to age-matched reference values from a large cohort of healthy participants (*n* = 1455, Reference Values for Arterial Stiffness, 2010) [17].

#### **3. Results**

Overall our data set included 60 (49 female) EDS participants age 13–70 years. There were no differences in age, height, weight, and body mass index between EDS participants of different types (Table 1).

#### *3.1. Pulse Wave Velocity in EDS*

Arterial elasticity did not differ by EDS type (Table 1). Grouped together, central PWV is lower in participants with EDS (4.73 ± 0.16 cm/s) compared to reference values in a large sample of healthy participants (Figure 1). PWV increases with age in healthy populations but the increase in arterial stiffness with aging is attenuated in people with EDS.

**Figure 1.** Pulse wave velocity (PWV) by age in participants with Ehlers-Danlos syndromes (EDS) compared to reference values in healthy humans' data from normal subjects in Reference Values for arterial Stiffness Collaboration (RVASC) [17] (*n* = 1455). Data are shown as mean ± standard deviation.

#### *3.2. Orthostatic Blood Pressure in EDS*

In the supine posture, BP and heart rate did not vary by EDS type (Table 1). In the standing posture, there was more variability in BP and heart rate measurements within each EDS type as shown by higher standard deviations compared to supine measurements demonstrating a wide range in responses to orthostasis (Table 1). Systolic BP in the standing posture was different between EDS types (ANOVA, *p* = 0.003). Post-hoc analysis showed that standing systolic BP was lower in participants with vascular EDS compared to those with hypermobile EDS (*p* = 0.021) and other/unspecified EDS (*p* = 0.002). Standing diastolic BP and heart rate also trended lower in the vascular EDS group (*p* = 0.087, Table 1.)

#### *3.3. Correlations between Pulse Wave Velocity and Blood Pressure*

Correlations of central and peripheral PWV to BP and heart rate are shown in Table 2. Central PWV did not correlate to HR or orthostatic BP changes over a 5 min period. Central PWV correlated significantly with supine (*r* = 0.387) and seated (*r* = 0.399) systolic BPs and supine diastolic BP (*r* = 0.400). Peripheral PWV did not correlate to HR or orthostatic BP. Peripheral PWV was correlated to diastolic BP in the supine (*r* = 0.322), seated (*r* = 0.383), and standing (*r* = 0.323) postures. All significant correlations were positive indicating that lower PWV (more elasticity) is associated with lower BP in our cohort of EDS participants.



57

Body mass index

 deviation.


**Table 2.** Correlations of pulse wave velocity to orthostatic hemodynamics in Ehlers-Danlos syndromes

Pearson's r-correlations are shown. Change in (Δ), Systolic blood pressure (SBP), diastolic blood pressure (DBP), heart rate (HR). \* Significant correlation at *p* ≤ 0.05 level.

#### **4. Discussion**

#### *4.1. Overall Findings*

This study used PWV to evaluate arterial stiffness in a diverse sample of people with different EDS types. This study provides three novel findings. We demonstrated that PWV is lower in people with EDS compared to reference values in the healthy population implying that their arteries are more elastic. We also found that lower PWV (indicating greater elasticity) is associated with lower systolic and diastolic BP in people with EDS. These findings may help explain the connection between EDS and impaired autonomic cardiovascular control. We also found no differences in PWV measurements among EDS types which suggests that the elasticity of the vasculature is similar among the diverse types of EDS.

#### *4.2. Significance of Decreased Pulse Wave Velocity in Ehlers-Danlos Syndrome*

The clinical association between EDS and orthostatic intolerance was identified in 1999 by Rowe et al. who first hypothesized that the mechanism connecting these two disorders is an increased enhanced elasticity in the arteries of people with EDS, predisposing them to OI [8]. Two decades later, this theory has become widely accepted despite the lack of empirical data to support it [3].

PWV has become the gold standard for assessing arterial structure because it is reproducible and aligns with more invasive measures. Its ease of use means it is available for testing in larger cohorts [9]. PWV has become a popular and validated method to assess increased stiffness of the central and peripheral vascular system in healthy humans and disease populations ranging from cardiovascular to neurological disease [9,10]. However, this technique is less commonly used to assess populations with increased arterial elasticity.

Three studies have previously measured PWV in people with EDS [11–13]. In a single family of 27 people with vascular EDS, Francois et al. utilized an older method for measuring pulse wave velocity involving piezo crystal microphones over the carotid, femoral, and dorsal arteries, and reported significantly decreased PWV (outside 2 standard deviations of normal values) in 5/27 participants studied [12]. A more recent study used the ultrafast ultrasound technique in 102 healthy participants and 37 vascular EDS participants and found that that central PWV was not significantly different in vascular EDS participants compared to controls [13]. Cheng, et al. employed a similar tonometry technique as was used in the current study to assess PWV in nine people with comorbid hypermobile EDS and POTS and nine age, sex, and BMI matched healthy controls, and found a trend to lower central PWV measurements in the people with EDS/POTS compared to controls [11]. Our study adds to the current literature by measuring both central and peripheral PWV in a larger and more heterogeneous group of people with EDS. In contrast to prior studies, we found that central PWV was significantly decreased in people with EDS compared to reference ranges for healthy subjects. This is likely due to our larger and more diverse sample.

The issue of age-associated changes in vascular function in people with vascular EDS was addressed by Mirault et al. using ultrafast ultrasound imaging (a method used by this group to measure PWV). They reported that the age-associated increase in vascular stiffness was attenuated in the vascular EDS participants [13]. We observed a similar phenomenon, namely that the PWV increased very little with progressive age deciles (Figure 1) which differs from reference values derived from healthy humans. High PWV (implying increased arterial stiffness) is related to adverse cardiovascular events in large epidemiology studies [18–20]. While one may speculate that lower PWV may be cardio-protective, it is unclear whether increased arterial elasticity is beneficial in people with EDS. Whether PWV has prognostic value in EDS deserves further investigation.

#### *4.3. Association between Pulse Wave Velocity and Blood Pressure*

Overall, lower PWV is related to lower BP measurements but is not directly indicative of orthostatic tolerance in EDS. These findings are consistent with measurements in healthy subjects and in other patient populations in which PWV tracks similarly to BP [10]. All significant correlations were positive indicating that lower PWV (more elasticity) is associated with lower BP in our cohort of people with EDS. We cannot infer causation from these data.

#### *4.4. Comparisons between EDS Types*

We did not see a difference in most orthostatic vital signs between EDS types. Systolic BP was slightly lower in vascular EDS which may reflect a difference in physiology or medications taken. Overall, there was a huge range in BP and heart rate responses to orthostasis which demonstrates inconsistent hemodynamic responses in this population and may reflect the presence of different types of OI. According to Roma, et al., about half of people with EDS have POTS (increase in heart rate of 30 beats/minute while standing) but others have orthostatic hypotension or hypertension [3]. It has been thought that vascular EDS was unique in terms of increased arterial distensibility. Our data are the first to compare arterial elasticity among EDS types in a single study, and demonstrate no difference in PWV among types of EDS. This is an important point, and it provides a possible explanation for the common presence of orthostatic intolerance in all EDS types.

#### *4.5. Strengths*

Strengths of this this study are the large sample size with a diverse EDS cohort including several EDS types, and the concomitant measurement of orthostatic vital signs and PWV. We compared central PWV measurements to published reference values in a large cohort. To our knowledge, there are no peripheral PWV reference values from large populations. The methods used in this study were consistent with study protocols used in the Baltimore Longitudinal Study of Aging [16].

#### *4.6. Limitations*

This study had several limitations. First, we did not include a contemporaneous control group in this study. However, we used reference values from a large cohort of healthy volunteers for comparison [16]. Second, participants were accessed while on medications which may impact BP, heart rate, and PWV assessments. Orthostatic vital signs were measured following 5 min in the supine posture then after 5 min sitting then 5 min of standing. This limits the ability to diagnose orthostatic intolerance as current diagnostic criteria for orthostatic intolerance involves hemodynamic measurements from the supine to standing posture after at least 10 min [21]. Finally, we acknowledge the heterogeneity of our EDS participants as a potential problem. Since these data were collected prior to the 2017 reclassification of EDS [1,14,15]. It is possible that some participants classified as having hypermobile or unspecified EDS would be categorized as hypermobility spectrum disorders using current criteria.

#### **5. Conclusions**

Overall, this is the first report of increased arterial elasticity in all types of EDS. The increased arterial elasticity was associated with lower supine and seated systolic and diastolic blood pressure in all types of EDS. We did not see differences in PWV in different types of EDS but standing systolic and diastolic blood pressure were lower in vascular EDS compared to the hypermobile and unspecified types. Our findings suggest that increased arterial elasticity may be related to impaired blood pressure control in EDS. Further studies are needed to determine whether this pathophysiological finding relates to orthostatic symptoms in people with EDS.

**Author Contributions:** Conceptualization: A.J.M., J.R.S., R.B., C.A.F.; Formal analysis: A.J.M., J.R.S., T.S.; Funding acquisition: J.R.S., R.B., C.A.F.; Supervision: J.R.S., R.B., C.A.F.; Writing—original draft: A.J.M., R.B., C.A.F.; Writing—review and editing: A.J.M., J.R.S., T.S., R.B., C.A.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Institute on Aging Intramural Research Program, under Protocol 2003-086 (later changed to 03-AG-N330) and by the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant UL1 TR002014. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This project was also supported by the Patient-Centered Outcomes Research Institute (PCORI) Eugene Washington Engagement Award (Title: The STRETCH Project: To Build Capacity Advancing Patient-centered Research in Ehlers-Danlos Syndrome).

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

#### **Abbreviations**


#### **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* **Cellular and Molecular Mechanisms in the Pathogenesis of Classical, Vascular, and Hypermobile Ehlers-Danlos Syndromes**

#### **Nicola Chiarelli, Marco Ritelli, Nicoletta Zoppi and Marina Colombi \***

Division of Biology and Genetics, Department of Molecular and Translational Medicine, University of Brescia, 25121 Brescia, Italy

**\*** Correspondence: marina.colombi@unibs.it

Received: 26 June 2019; Accepted: 9 August 2019; Published: 12 August 2019

**Abstract:** The Ehlers-Danlos syndromes (EDS) constitute a heterogenous group of connective tissue disorders characterized by joint hypermobility, skin abnormalities, and vascular fragility. The latest nosology recognizes 13 types caused by pathogenic variants in genes encoding collagens and other molecules involved in collagen processing and extracellular matrix (ECM) biology. Classical (cEDS), vascular (vEDS), and hypermobile (hEDS) EDS are the most frequent types. cEDS and vEDS are caused respectively by defects in collagen V and collagen III, whereas the molecular basis of hEDS is unknown. For these disorders, the molecular pathology remains poorly studied. Herein, we review, expand, and compare our previous transcriptome and protein studies on dermal fibroblasts from cEDS, vEDS, and hEDS patients, offering insights and perspectives in their molecular mechanisms. These cells, though sharing a pathological ECM remodeling, show differences in the underlying pathomechanisms. In cEDS and vEDS fibroblasts, key processes such as collagen biosynthesis/processing, protein folding quality control, endoplasmic reticulum homeostasis, autophagy, and wound healing are perturbed. In hEDS cells, gene expression changes related to cell-matrix interactions, inflammatory/pain responses, and acquisition of an in vitro pro-inflammatory myofibroblast-like phenotype may contribute to the complex pathogenesis of the disorder. Finally, emerging findings from miRNA profiling of hEDS fibroblasts are discussed to add some novel biological aspects about hEDS etiopathogenesis.

**Keywords:** autophagy; collagen III; collagen V; Ehlers-Danlos syndrome; endoplasmic reticulum; extracellular matrix; fibroblast-to-myofibroblast transition; miRNA; transcriptome; wound healing

#### **1. The Extracellular Matrix: An Overview**

Connective tissues have an extracellular matrix (ECM) with a specific composition generated during embryogenesis and maintained in adult life. The ECM is a complex network that provides a structural scaffold to the surrounding cells and is a reservoir of bioactive molecules such as cytokines and growth factors that control cell behavior [1]. The main ECM components include proteoglycans, hyaluronic acid, adhesive glycoproteins such as fibronectin and laminins, and fibrous proteins like collagens and elastin [2]. Matricellular proteins such as thrombospondins, osteopontin, periostin, and tenascins are non-structural ECM proteins, primarily acting as mediators of cell–matrix interactions, which are abundantly expressed during embryonic development, wound healing, and tissues renewal [3].

The humanmatrisome consists of about 300macromolecules comprising the "corematrisome", whichis composed of many different collagens, proteoglycans (e.g., aggrecan, versican, perlecan, and decorin), and glycoproteins (e.g., laminins, elastin, fibronectin, thrombospondins, and tenascins) [4]. Matrisome also includes many matrisome-associated proteins and ECM-regulators, i.e., ECM-cross-linking (e.g., lysyl oxidases, transglutaminases) and ECM-modifying enzymes (e.g., proteases and their inhibitors) together with secreted factors including transforming growth factor β (TGFβ), wingless integrated (Wnt), and multiple cytokines [5].

Collagens represent the major ECM structural components and play a central role in providing the structural integrity of several connective tissues (e.g., cartilage and bone) and various organ systems including skin, lungs, blood vessels, and cornea. Collagens are also involved in cell adhesion, chemotaxis, and migration [1,2]. The dynamic interplay between cells and collagens regulates tissue remodeling during growth, differentiation, morphogenesis, and wound healing [6,7]. The common molecular feature of collagens is their triple helical structure, which consists of three collagenous α-chains with the typical recurring (Gly-Xaa-Yaa)n tripeptide sequence. The presence of glycine residues in the collagenous domain is essential for stability and correct assembly of the triple helix. Collagen biosynthesis, assembly, and maturation require a sequence of well-controlled intracellular and extracellular events (for review see [8,9]). Collagen I is the most abundant type expressed in bone, cornea, dermis, and tendon. Collagen III is primarily present in the tunica media of the blood vessels and hollow organs (e.g., uterus, intestine). Collagen V is widely distributed, especially in dermis, tendons, and muscles, playing a central role in collagen I fibrillogenesis [9]. These fibrillar collagens form structures necessary to ensure the strength and structural integrity of the ECM of all connective tissues and organs of the body [10].

Fibronectin is a dimeric and fibrillar glycoprotein ubiquitously organized in the ECM of all tissues and is also present in soluble form in the plasma. Cellular fibronectin self-assembles in fibers and binds collagens, fibrin, proteoglycans, and cell surface receptors, providing cell growth, adhesion, and migration. During wound healing, it forms a provisional matrix with fibrin and enters in the granulation tissue formation in the late phase of re-epithelization [11].

In addition to ensuring physical support and structural integrity, the proper ECM composition and organization are crucial for cell health. ECM undergoes a continuous turnover either under physiological or in pathological circumstances, and its homeostasis is critical for connective tissues architecture and function [12,13]. Integrins are specific cell surface receptors that mediate the complex cell-matrix interactions. These bridging molecules, which are heterodimeric transmembrane receptors containing α and β subunits, connect ECM to cytoskeleton by interacting via their extracellular domain with collagens and other matrix molecules and via their cytoplasmic tails with cytoskeleton components (e.g., actin, vinculin, talin, paxillin), thus mediating cell adhesion and motility [14,15].

#### **2. Pathological ECM Remodeling and Perturbation of Cellular Homeostasis**

Cell-matrix interaction via integrins is crucial for cell survival and tissue homeostasis. Prolonged loss of integrin-mediated cell–ECM adhesion leads to anoikis [16]. Under physiological conditions, ECM detachment triggers anti-apoptotic signals as a cell survival mechanism to delay the onset of anoikis. One of such signaling pathways is autophagy, which is a highly conserved cellular catabolic process that promotes homeostasis and mitigates the stress due to ECM detachment [17]. Autophagy is essential for cellular maintenance and homeostasis by promoting the turnover of macromolecules and organelles via the lysosomal degradative pathway [18]. Physiological and pathological changes in the ECM composition play a crucial role in modulating autophagy activity [19,20]. For instance, deficiency of collagen VI, which is associated with a spectrum of different myopathic conditions, perturbs ECM architecture, impairs the autophagic flux, and activates pro-apoptotic signals [21]. Autophagy, in turn, contributes to the maintenance of endoplasmic reticulum (ER) function by mediating its turnover through the autophagic sequestration of ER fragments into autophagosomes, the so-called ER-phagy process [22].

ECM components also modulate immune cell migration into inflamed tissues and their activation and proliferation [23]. The stimulation of the innate immunity results from the recognition of specific mediators, namely pattern recognition receptors, which, in turn, recognize molecules, referred as danger-associated molecular patterns, which are released from damaged tissues [24]. It is well documented that different ECM components or their fragments including the fibronectin 1 extra domain A, one of the alternative spliced regions of fibronectin encoding gene, tenascin-C, fibrinogen, and several proteoglycans, serve as danger signals and trigger immune responses following tissue damage or in response to pathological ECM remodeling [24,25]. In fibrotic conditions, increased ECM production, accumulation of ECM fragments, augmented secretion of cytokines, fibroblast-to-myofibroblast transition, and activation of immune responses, dependent on toll-like receptors, occur [26]. The regulation of ECM synthesis and remodeling is central for human health, as recognized in different heritable connective tissue disorders [1,27]. Indeed, molecular defects in a large range of ECM-related genes, including those encoding enzymes involved in biosynthesis or processing of ECM proteins, cause a myriad of connective tissue disorders, e.g., Ehlers–Danlos syndromes, Osteogenesis imperfecta, Marfan syndrome, Loeys–Dietz syndromes, arterial tortuosity syndrome, and numerous skeletal dysplasias [1,27]. These disorders are characterized by a multisystem involvement in terms of cardiovascular, skeletal, and cutaneous features [28], highlighting the functional relevance of the ECM in ensuring the integrity and function of several connective tissues.

The pathological consequences of defects in ECM components depend on the balance between extracellular effects, e.g., reduced protein secretion and export of misfolded proteins, and intracellular consequences such as apoptosis activation, ER dysfunction, and autophagy perturbation that impact in different ways on the molecular pathology and disease severity [27,29–31].

#### **3. Ehlers-Danlos Syndromes**

Ehlers-Danlos syndromes (EDS) represent a clinically and genetically heterogeneous group of conditions that share a variable combination of skin hyperextensibility, joint hypermobility, and internal organ and vessel fragility [32]. The 2017 international classification of the Ehlers-Danlos syndromes recognizes 13 subtypes, which are caused by pathogenic variants in 19 different genes, mainly encoding fibrillar collagens and collagens-modifying proteins [32]. EDS types are grouped based on the underlying genetic and pathogenetic mechanisms in disorders related to (i) collagens primary structure and processing (*COL1A1*, *COL1A2*, *COL3A1*, *COL5A1*, *COL5A2* and *ADAMTS2*), (ii) collagens folding and cross-linking (*PLOD1* and *FKBP14*), (iii) structure and function of the myomatrix, i.e., the specialized ECM of muscle (*TNXB* and *COL12A1*), (iv) glycosaminoglycans biosynthesis (*B4GALT7*, *B3GALT6*, *CHST14*, and *DSE*), (v) complement pathway (*C1S* and *C1R*), and (vi) intracellular processes (*SLC39A13*, *ZNF469*, and *PRDM5*). The classical (cEDS), vascular (vEDS) and the molecularly unsolved hypermobile (hEDS) EDS forms account for more than 90% of patients. Recently, a new and very rare EDS variant has been identified that is caused by biallelic mutations in the *AEBP1* gene (Table 1) [33–36].

The new nosology proposed for each subtype a set of major, minor, and minimal criteria addressing clinical suspicion for a specific EDS type and confirmatory molecular testing. For a comprehensive overview of all EDS forms see the landmark work by Malfait and colleagues [32].

The decrease in the tensile strength and integrity of skin, joints, and hollow organs is a common disease mechanism shared by the different EDS types [37]. This mechanical weakness is considered the driving factor of connective tissue fragility, even if it is likely that multiple cell-matrix interplays and involvement of distinct intracellular signaling pathways contribute to the molecular pathology of the different EDS phenotypes [38].

In the following chapters, we will review and expand the results derived from our previous transcriptome and in vitro studies on cEDS, vEDS, and hEDS patients' dermal fibroblasts. Taken together, these studies highlighted that the alteration of the ECM structural integrity is a common disease factor contributing to the pathogenesis of all these conditions.


**Table 1.** EDS types grouped according to the underlying genetic defect and pathomechanisms.

<sup>A</sup> New EDS variant recently defined in [33–36]. AD: autosomal dominant; AR: autosomal recessive; IP: inheritance pattern.

In cEDS and vEDS fibroblasts, the ECM disarray is a direct consequence of molecular defects in respectively collagen V and collagen III that impair common molecular functions essential to guarantee adequate folding and maturation of proteins and biological processes crucial for cell survival and homeostasis. The ECM disorganization observed in hEDS cells may be a consequence of an excessive pathological turnover, mainly due to ECM-degrading enzymes and other so far unknown factors, which might be primary contributors involved in the transition to a pro-inflammatory myofibroblast-like phenotype. Consistently, the perturbation of distinct transcriptional patterns observed in cEDS, vEDS, and hEDS fibroblasts pointed out different disease mechanisms underlying the pathophysiology of these EDS cell types.

Altogether, these insights represent a starting point for future investigations on the numerous pathobiological aspects underlying these conditions. An overview of the biological findings emerged from transcriptome and in vitro studies on dermal fibroblasts from cEDS, vEDS, and hEDS patients is summarized in Table 2.


**Table 2.** Overview of the biological processes dysregulated in cEDS, vEDS, and hEDS patients' dermal fibroblasts emerged from transcriptome and in vitro studies.

+: detected by transcriptome or in vitro studies, −: not experimentally detected by transcriptome or in vitro studies, na: not ascertained.

#### **4. Classical Ehlers-Danlos Syndrome**

Classical EDS (cEDS, OMIM#130000) is characterized by marked skin involvement, generalized joint hypermobility, and abnormal wound healing [32,39]. Most patients harbor point mutations or chromosomal rearrangements in *COL5A1* or *COL5A2* genes encoding the collagen V [40,41]. This collagen is abundantly distributed in a variety of tissues as heterotrimers, which co-assemble with collagen I to form heterotypic fibrils [42].

Collagen V knockout mice synthesize and secrete normal amounts of collagen I, but collagen fibrils are absent, and the animals die at the onset of organogenesis, supporting the crucial role of collagen V for embryonic development [43].

Collagen V haploinsufficiency is the most common molecular defect caused by *COL5A1* null alleles, whereas rare *COL5A1* variants and the majority of *COL5A2* mutations reported so far affect collagen V structural integrity by exerting a dominant negative effect [40,41].

#### **5. Altered ECM Turnover, Wound Healing, and Inflammation in cEDS Fibroblasts**

Although the reduced availability of collagen V is crucial in the pathogenesis of cEDS, the molecular aspects contributing to the pathophysiology of the disorder remain poorly characterized. Our in vitro findings demonstrated that cEDS patients' fibroblasts show disassembly of many ECM components, including collagen V and III, fibronectin, and fibrillins, and disorganization of collagen- and fibronectin-specific α2β1 and α5β1 integrin receptors [44–47]. cEDS cells also exhibit a reduced in vitro migration capability, an abnormal wound healing response, and a crosstalk involving the αvβ3 integrin and epidermal growth factor (EGF) receptor that rescues them from anoikis [44–50]. In line with these in vitro findings, *Col5a1* and *Col5a2* deficient mice show a defective wound healing response and reduced cell migration [51,52].

Transcriptome profiling of cEDS fibroblasts added new insights into the complex molecular mechanisms involved in the maintenance of ECM homeostasis and proper wound healing, since patients' cells showed the dysregulated expression of many genes encoding matricellular and soluble proteins with prominent functions in cell proliferation and migration, collagen assembly and ECM remodeling during wound healing, i.e., *SPP1*, *POSTN*, *EDIL3*, *IGFBP2*, and *C3* [47]. Wound healing is a highly controlled multistep process involving several growth factors, cytokines, matrix metalloproteases, and cellular receptors, as well as proper crosstalk of different ECM constituents essential for ensuring

tissue regeneration [53]. In addition to ECM glycoproteins and collagens, many other matricellular proteins including osteopontin, periostin, and tenascins are required for the formation of a provisional ECM during wound repair [54,55].

Osteopontin encoded by *SPP1,* which shows a decreased expression in patients' cells, is involved in several physiological processes related to inflammation, biomineralization, cell viability, and wound healing [56,57]. Through its interaction with the αvβ3 integrin, osteopontin facilitates the adhesion of bone cells during bone tissue formation by stimulating a mineralized collagen ECM [58,59]. Finally, the functional role of osteopontin in the ECM reorganization during wound healing is crucial, since *Opn*-deficient mice show ECM disorganization and disassembly of collagen fibrils in the deep layers of wound sites [60].

In cEDS fibroblasts, the decreased expression of periostin encoded by *POSTN* may also contribute to the generalized ECM disarray and in vitro poor wound healing [44,48]. Indeed, periostin plays an important role in ECM structure and organization and particularly in collagen assembly, by acting as a scaffold protein for the bone morphogenetic protein 1, which facilities the proteolytic activation of lysyl oxidase that, in turn, catalyzes the covalent cross-link formation of collagens [61]. Consistently, *Postn*-deficient mice exhibit marked reduction of collagen cross-linking and increased levels of collagen fragments owing to proteolytic digestion [62,63]. Periostin also acts as a pro-survival protein in many cellular circumstances by interacting with αvβ3 and αvβ5 integrin receptors and mediating the activation of several intracellular signaling pathways [64]. In wound sites, it promotes activation of fibroblasts during wound contraction and stimulates collagen assembly and ECM reorganization [65,66]. *POSTN* not only shows a diminished expression in vEDS cells but also in dermal fibroblasts from patients with *FKBP14*-kEDS [67], further emphasizing the crucial role of periostin as a scaffold matricellular protein necessary for collagen assembly and ECM stability.

*EDIL3* (EGF-like repeat- and discoidin I-like domain-containing protein 3), the most down-regulated transcript in cEDS fibroblasts, encodes an ECM-associated protein that promotes angiogenesis in vitro through binding to αvβ3 and αvβ5 integrins [68]. It stimulates cell migration and proliferation, mediates apoptotic cell phagocytosis, regulates neutrophil recruitment to the inflamed tissue, and prevents chondrocyte anoikis through its interaction with the αvβ3 integrin [69–71].

*IGFBP2* (insulin-like growth factor-binding protein 2), the most up-regulated gene in cEDS cells, enhances cell migration in different cell types through its binding to αvβ3 and α5β1 integrins [72,73]. Its high expression in cEDS cells might represent a transcriptional response in the attempt to counteract, at least in vitro, their reduced migration capability [48,49].

Of note is also the marked up-regulation of the complement factor C3 belonging to a complex network of plasma and membrane proteins involved in the innate immunity [74]. Complement can modulate the inflammatory response during wound healing to restore tissue injury; however, its unbalanced or prolonged activation can exacerbate inflammation, delaying the physiological wound healing [75]. Specifically, C3 functions as a negative regulator of tissue healing, since *C3*-deficient mice exhibit an increased wound healing and angiogenesis [76].

Taken together, these gene expression abnormalities expand the current understanding of altered molecular mechanisms underlying the deficient wound healing response observed in cEDS cells. Additional functional work might help to establish the concrete involvement of these proteins including the αvβ3 integrin in the impaired wound healing, which likely leads to the cutaneous manifestations of cEDS [39,77].

#### **6. Perturbation of ER Homeostasis and Autophagy in cEDS Fibroblasts**

ER is a fundamental cellular organelle involved in the maintenance of numerous aspects of cell health by ensuring folding and exporting of secretory or transmembrane proteins [78]. The biosynthesis, processing, and integrity of collagens and other ECM structural constituents are critical for intracellular proteostasis [79]. To restore intracellular equilibrium, the ER counteracts the accumulation of aggregated

or misfolded proteins by means of quality control mechanisms such as unfolded protein response, ER-associated degradation, and autophagy [80].

In cEDS fibroblasts, a possible unbalance of ER homeostasis and autophagy was assumed given the decreased expression of many associated genes such as *DNAJB7*, *ATG10*, *CCPG1*, and *SVIP* [47]. *DNAJB7* encodes a member of the J protein/heat shock protein family acting as ER chaperones in the quality control of aggregate protein [81]. *ATG10* is a member of the autophagy related proteins family that participates to the generation and expansion of autophagosomes [22]. The protein encoded by *CCPG1* acts as an ER-phagy cargo receptor facilitating the attachment to growing autophagosomes of the microtubule-associated LC3 protein that is crucial for autophagosome maturation [22,78,82].

CCPG1 plays a key role in the ER proteostasis, since its deficiency causes accumulated insoluble proteins and consequent ER dilation [83]. A recent study reported the contribution of ER-phagy in the selective degradation of misfolded procollagen I molecules via a calnexin-FAM134B complex [84]. Furthermore, inefficient procollagen folding in the ER may induce autophagy as a cytoprotective mechanism [85]. Based on these findings, it is reasonable to speculate a possible role of CCPG1 in this autophagy-dependent mechanism and that its decreased expression in cEDS fibroblasts might impair the ER quality control. The disturbance of ER homeostasis in cEDS is also suggested by the decreased transcription of the small VCP/p97-interacting protein (SVIP), which is a modulator of the ER-associated degradation pathway [86,87]. Previous and recent data highlighted the contribution of SVIP in the regulation of autophagy, since its overexpression is associated with increased LC3 lipidation and attenuation of hepatic fibrosis by the induction of the autophagic flux [88,89].

The impairment of ECM organization, matrix-cell interactions, and the activation of ECM-dependent intracellular signaling may elicit autophagy [18,19]. Cell detachment from the ECM activates the autophagy pathway that, in turn, protects cells from anoikis [90]. Moreover, depletion of autophagy regulators is associated with induction of pro-apoptotic signals, decrease of collagen degradation via lysosome pathway, and regulation of cell adhesion [90,91]. In line with these observations, the aberrant expression of collagen V and defective remodeling of ECM in cEDS cells might affect ER homeostasis and autophagy, and consequently activate a pro-survival mechanism mediated by a crosstalk between αvβ3 integrin and EGF receptor [45,50]. Additional studies are needed to better elucidate the contribution of these processes in the molecular pathology of cEDS.

#### **7. Vascular Ehlers-Danlos Syndrome**

Among the different EDS forms, vascular EDS (vEDS, OMIM#130050) is the most severe type and is primarily characterized by life-threatening features of tissue fragility leading to arterial dissection or aneurysm, gastrointestinal ruptures, and pregnancy complications at a young age [92,93].

vEDS is caused by mutations in *COL3A1* encoding collagen III that shows a predominant expression in blood vessels and hollow organs [94]. Most disease-causing variants in *COL3A1* are glycine substitutions that destroy the triple helical winding, thus altering the structural integrity of collagen III due to misfolded procollagen III in the ER, and thereby impairing the secretion and deposition into the ECM of functionally mature molecules [95,96].

*Col3a1* deficient mice show a reduced lifespan mainly due to arterial ruptures and abnormalities of collagen fibril organization in several collagen-rich organs, i.e., aorta, skin, lung, and bowel [10].

Our previous protein findings on cultured patients' fibroblasts showed that dominant negative mutations in *COL3A1* lead to the reduced secretion of collagen I into the ECM [44], consistent with the known regulatory role of collagen III in synthesis and deposition of heterotypic fibrils that largely contain collagen I [96].

#### **8. Disturbance of ECM Organization, Collagens Processing, and ER Homeostasis in vEDS Fibroblasts**

Although it is well known that the disruption of the collagen III triple helical structure leads to abnormal protein folding, different biological aspects of the vEDS pathogenesis are not yet fully studied. In line with cEDS and hEDS transcriptome profiling, vEDS fibroblasts show the differential

expression of several genes encoding structural constituents of the ECM, further supporting the notion that abnormal ECM remodeling is a common denominator of these conditions [50,97].

vEDS cells showed a marked decrease in expression of the fibrillin 2 encoding gene (*FBN2*). Fibrillins are essential structural ECM components involved in the organization of blood vessels and dermis, and in combination with elastic fibers they act as scaffold to ensure tissue elasticity [98]. Moreover, fibrillins interact with elastin microfibril interface-located proteins (EMILINs) and facilitate their incorporation into the dermal ECM [99].

Fibrillins also regulate the bioavailability of the TFGβ through the interaction with latent TFGβ binding proteins [99]. Specifically, fibrillin 2 plays a role in bone and soft connective tissue morphology by influencing the collagen cross-linking [100].

Besides its role in elastogenesis and ECM stability, fibrillin 2 also has a role during wound healing [101]. About this, vEDS fibroblasts share with cEDS cells the decreased expression of the related periostin-encoding gene and show reduced migration capability [49]. Consistently, both altered wound healing and reduced total collagen content were reported in a *Col3a1* transgenic mouse model [102].

Our protein findings confirmed the pathological ECM remodeling of vEDS fibroblasts, as a generalized fibrillin disarray in combination with the disassembly of EMILINs and elastin network was revealed, consistent with the extreme vascular fragility observed in vEDS patients [97]. The disorganization of core proteins of the proteoglycans perlecan, versican, and decorin, which are involved in the formation of collagen fibrils, further emphasizes the widespread ECM disarray and altered collagens biosynthesis/secretion of vEDS cells that are consequent to collagen III defect [97].

The biosynthetic pathway of fibrillar collagens is a highly regulated process involving folding enzymes, molecular chaperones, and post-translational modifications essential for proper protein assembly, stability of collagen fibrils, and their transport to the cell surface [8,9]. In vEDS cells, this complex machinery seems to be perturbed given the reduced expression of many ER-resident enzymes involved in different steps of collagen biosynthesis, i.e., *P4HA2, P4HA3, LOXL3,* and *FKBP14*.

*P4HA2* and *P4HA3* encode the α-subunit of the collagen prolyl-4-hydroxylase, which catalyzes the hydroxylation of collagen prolyl residues necessary to provide thermal stability to the collagen triple helix. Lysyl oxidase-like 3 (*LOXL3*) stabilizes the formation of intra- and intermolecular crosslinks during assembly of collagen and elastin fibrils. *FKBP14* encodes a peptidyl-prolyl cis-trans isomerase (FKBP22) that catalyzes in the ER lumen collagen folding and it functions as a molecular chaperone for different collagens including collagen III [8,103]. Dermal fibroblasts of *FKBP14*-deficient patients show a generalized perturbation of protein folding and a consequent enlargement of ER cisternae [104]. The marked decrease of FKBP22 protein levels observed in vEDS cells suggests an ER accumulation of misfolded proteins, consistent with the possible dilation of ER cisternae evinced by immunofluorescence analysis with the ER marker PDI [97].

Structural mutations in different collagen types disturb the assembly into hetero- or homotrimers or lead to abnormal triple helix folding. The consequent accumulation of misfolded collagen molecules into the ER lumen activates the proteasomal degradation system to re-establish ER proteostasis [27]. In vEDS cells, this quality control machinery might not work properly, given the decreased transcription of several genes encoding different catalytic and non-catalytic subunits of the proteasome complex, such as *PSMA6*, *PSMB6*, *PSMC3*, and *PSMD2*. In addition, the reduced transcription of members belonging to the DnaJ heat shock protein family, i.e., *DNAJB7*, *DNAJB11*, *DNAJC3, DNAJC10*, and *DNAJC24*, and to the thioredoxin superfamily, i.e., *TXN*, *PDIA4*, *PDIA5*, and *PDIA6*, which all act as intracellular mediators for correct protein folding and intracellular redox balance [105], further corroborates a perturbed ER proteostasis in vEDS fibroblasts.

This imbalance can be overcome by the activation of stress-related pathways, such as unfolded protein response and autophagy to restore basal cellular equilibrium [80]. The alteration of the ER redox state may also trigger pro-death signals through the regulation of members of the Bcl-2 family and activation of caspase-dependent apoptosis [106,107]. Consistently, we previously demonstrated that vEDS fibroblasts are in a pre-apoptotic state, due to downregulation of the Bcl-2 anti-apoptotic

protein and increased levels of caspase enzymes, and activate a cell survival mechanism through an αvβ3-EGFR crosstalk [45,50]. However, in vEDS cells an enhanced expression of unfolded protein response-related genes was not identified, consistent with the recent findings on cultured dermal fibroblasts from *Col3a1* transgenic mice that did not show elevated levels of the unfolded protein response markers *Bip* and *Chop* [102]. In line with this evidence, a recent transcriptome analysis of dermal fibroblasts from *FKBP14*-kEDS patients did not reveal a high expression of genes associated with ER stress and unfolded protein response activation [67], though early data on this EDS cell type suggested an enlargement of ER [104]. Nevertheless, given that different reports highlighted the role of ER stress in the pathogenesis of several collagenopathies [27,108,109], further work is warranted to explore the possible ER perturbation as a disease mechanism of vEDS to identify novel potential therapeutic targets.

#### **9. Hypermobile Ehlers-Danlos Syndrome**

Hypermobile EDS (hEDS, OMIM#130020), mainly characterized by generalized joint hypermobility and its complications, minor skin changes, and apparently segregating with an autosomal dominant pattern, is still without a known molecular basis. The phenotypic spectrum of hEDS is wide-ranging and heterogeneous and further complicated by multiple associated symptoms shared with other heritable or acquired (autoimmune) connective tissue disorders and chronic inflammatory systemic diseases [32,110,111].

Despite the significant advances in molecular genetic techniques, attempts to disclose the genetic cause(s) of hEDS have been so far inconclusive. Several studies struggled to define its genetic etiology but without compelling evidence, corroborating the hypothesis of a high genetic heterogeneity of the condition [112–115]. The introduction of more selective clinical criteria for hEDS in the novel classification aimed to minimize heterogeneity allows for the formation of homogeneous cohorts to facilitate scientific research to discover the underlying genetic cause(s) of the condition [32]. Nowadays, hEDS is considered at one end of a continuous spectrum of phenotypes, which originates from isolated non-syndromic joint hypermobility and passing through the recently defined hypermobility spectrum disorders (HSD) [111]. HSD refers to patients who present symptomatic joint hypermobility but do not fulfill the new diagnostic criteria of hEDS. Recently, given the clinical continuity between hEDS and HSD and our data on patients' dermal fibroblasts [49], it was proposed that these disorders might be considered as a single entity, referred to as hEDS/HSD [116], as already occurred for the hypermobility type of EDS and joint hypermobility syndrome [117]. Until now, no validated biological biomarkers have been identified for recognizing hEDS/HSD, which are dominated by extremely variable phenotypes and chronic disability affecting patients' quality of life [111,116].

In this intricate scenario, the integration of various biological knowledge could be an effective strategy to delineate molecular mechanisms contributing to the disease pathophysiology. Transcriptome and proteome profiling can be useful to reveal specific biological signatures, thus providing insights not only for the understanding of the pathomechanisms but also for the identification of reliable tools for therapeutic options [118–122].

Previous findings based on transcriptome and protein studies on a cohort of hEDS/HSD patient-derived dermal fibroblasts represent up to now the main effort to unravel their complex etiopathogenesis [48,123]. Proteome profiling of patients' cells is currently ongoing to corroborate these data, since the gene expression profiling and cellular studies on patients' fibroblasts provided significant clues that are likely relevant for the disease pathogenesis. In the following paragraphs we review our past findings and discuss some novel emerging aspects, offering future perspectives for molecular research in this field.

#### **10. Pathological ECM Remodeling and Defective Cell-Cell Interactions in hEDS**/**HSD Cells**

Although hEDS/HSD etiology remains elusive, patients' skin fibroblasts show a disorganization of the ECM like that observed in cells derived from the other EDS types. In particular, hEDS/HSD, cEDS, and vEDS fibroblasts exhibit a marked disorganization of collagen and fibronectin ECM and their specific α2β1 and α5β1 integrin receptors and showed the preferential expression of the αvβ3 integrin [44,49,50,123].

Transcriptome of hEDS/HSD fibroblasts revealed the dysregulated expression of several genes encoding either ECM glycoproteins such as elastin (*ELN*) and sparc/osteonectin (*SPOCK*), ECM regulators, i.e., metalloproteinases (*MMP16*, *PAPPA2*) and transglutaminase (*TGM2*), or ECM associated secreted factors such as secreted frizzled-related protein 2 (*SFRP2*) and transforming growth factor alpha (*TGFA*). This ECM signature is in common with cEDS and vEDS cells, underlining that the matrix perturbation may act as a key driving factor for the EDS pathogenesis, irrespective of the underlying molecular defects [50].

Cells sense the intrinsic mechanical properties of the ECM and convert these stimuli into intracellular signals [124]. In addition to integrins that primarily mediate this cell response, intracellular signals may be triggered also through the cadherin superfamily, which are calcium-dependent transmembrane proteins forming complex adhesions and connect to the actin cytoskeleton via numerous proteins [125,126]. Interestingly, transcriptome profiling of hEDS/HSD cells revealed a differential expression of many adhesion molecule-encoding genes including members of cadherins and protocadherins, i.e., *CDH2*, *CDH10*, *PCDH9*, *PCDHB16*, *PCDHB18*, claudins (*CLDN11*), and desmosomes (desmoplakin, *DSP*), which are involved in the formation of specialized cell-cell junction complexes essential for maintaining epithelial integrity, morphogenesis, and tissue architecture [127,128]. Since these adhesion proteins can act as signaling modulators of intracellular pathways, such as Wnt, Hippo, NF-kB, JAK-STAT that are crucial for development and organogenesis [128], their altered expression could impact on multiple biological processes essential for embryogenesis and tissue homeostasis.

These transcriptional changes suggested a fibroblast-to-myofibroblast transition of hEDS/HSD cells. This phenomenon induces the formation of cells with muscle-like features that are characterized by increased cell contractility, formation of alpha smooth muscle actin (α-SMA)-stress fibers, together with the reorganization of cell-matrix and cell-cell contacts and cytoskeletal architecture [127,129–131]. Our in vitro studies confirmed the phenotypic conversion of hEDS/HSD fibroblasts into migrating myofibroblast-like cells, since they express the typical markers α-SMA and cadherin-11 and show augmented levels of the protease MMP9 and an altered expression of the inflammation mediators CYR61 and CTGF [49]. This phenotypic switch is elicited by a signal transduction pathway involving the αvβ3 integrin that signals through the integrin linked kinase (ILK) and the transcription factor Snail1 [49,50]. This myofibroblast-like phenotype observed in vitro might reflect a persistent in vivo inflammatory-like condition consistent with the patients' systemic clinical manifestations, comprising gastrointestinal dysfunction, increased susceptibility to osteoarthritis, chronic generalized musculoskeletal pain, inflammatory soft-tissue lesions, and neurological features [110,111].

Activation of myofibroblasts is itself part of physiological wound repair following tissue injury, whereas in chronic injury and inflammatory fibrotic conditions their persistent activation exacerbates the disease progression [132–135]. During fibroblast-to-myofibroblast transition, a complex mechanochemical signaling is activated involving profibrotic secreted factors such as TGFβ and Wnt and ECM-degrading enzymes and intracellular effectors required for stress fiber contractility [127,131]. At molecular level, the cytokine TGFβ is considered the master regulator of profibrotic processes. A growing body of evidence has highlighted the regulation of the Wnt/β-catenin pathway by TGFβ as well as the involvement of their downstream molecular effectors in the fibroblast-to-myofibroblast transition and fibrotic responses [136].

As revealed by transcriptome analysis, several signaling pathways essential for cell growth and proliferation were found to be likely perturbed in hEDS/HSD fibroblasts, i.e., TGFβ, TNF, Jak-STAT, and PI3K-Akt [123]. Transcriptomics data also suggested the differential expression of different Wnt-related genes including the up-regulated frizzled receptor 3 (*FZD3*) and the down-regulated Wnt negative regulators *PRICKLE1* and *SFRP2*. *SFRP2*, the most down-regulated transcript in patients' cells, acts as a critical Wnt modulator, since it directly binds to Wnt proteins and prevents their interactions with FZD receptors [137].

The synergistic crosstalk between TGFβ and Wnt signaling in the myofibroblast activation is documented as well as the inhibitory role of SFRP2 in the TGFβ-dependent myofibroblast formation and post-inflammatory fibrosis [138,139]. In this view, a possible involvement of these signaling pathways in the pathomechanisms of hEDS/HSD can be envisaged. Our findings may offer further clues to address important questions concerning the activation of these pathological mechanisms, and it remains to be clarified which growth factors, i.e., TGFβ, CTGF, and key regulatory pathways sustain the fibroblast-to-myofibroblast transition of hEDS/HSD cells.

#### **11. Di**ff**erential Expression of Genes Involved in Inflammatory, Immune, and Pain Response in hEDS**/**HSD Cells**

Over the past few years, clinical research described the presence of comorbidities in hEDS/HSD patients, such as functional gastrointestinal and eosinophilic disorders [140,141], increased prevalence of asthma [142], and chronic pain syndromes, i.e., chronic fatigue, fibromyalgia, irritable bowel disease, and inflammatory joints conditions [111,143], though specific underlying causes and mechanisms remain to be explored. In this regard, transcriptome of hEDS/HSD cells revealed the aberrant transcription of a range of genes related to inflammation, pain, and immune responses, i.e., *AQP9*, *CFD*, *SPON2*, *PRLR*, and *NR4A* receptors, which might impair biological functions and molecular pathways with a potential role in the disease's pathogenesis. Among them, patients' cells showed the enhanced expression of *AQP9*, a member of the family of water-selective membrane channels that play a role both in antimicrobial defense and skin barrier permeability [144]. A high expression of this transporter was detected in synovial tissues and fibroblast-like synoviocytes from osteoarthritis and rheumatoid arthritis patients and may have a role in the pathogenesis of inflammatory synovitis [145,146].

In line with this finding, patients' cells also showed increased mRNA levels of complement factor D (*CFD*), a component of the alternative complement pathway [147], which is involved in pathophysiological mechanisms of osteoarthritis and is considered as a potential predictive biomarker of joint pain in patients with hip and knee osteoarthritis [148–150].

In hEDS/HSD no reliable biomarkers have been identified. A previous study identified elevated basal serum tryptase levels due to increased *TPSAB1* copy number associated with hereditary alpha tryptasemia in individuals with multisystem complaints, i.e., joint hypermobility, sleep disruption, irritable bowel syndrome, body pain, headache, arthralgia, and chronic gastroesophageal reflux, partly overlapping with those frequently observed in hEDS/HSD patients [151]. In our hEDS/HSD patients, no elevated basal serum tryptase level was observed, suggesting the absence of the association between their clinical features and copy number variations in the *TPSAB1* gene.

Other inflammation-related genes dysregulated in hEDS/HSD include *SPON2*, up-regulated in patients' cells, which encodes an ECM protein with multifunctional properties in the innate immune system and inflammatory cell recruitment [152–154], and *PRLR*, showing a decreased expression in hEDS/HSD cells, which encode the prolactin receptor implicated in inflammatory responses and immune cells regulation [155,156]. The prolactin-PRLR axis contributes to the activation of pain-related pathways through the sensitization of transient receptor potential channels that promote painful sensations [157,158]. Chronic pain represents a common complaint among hEDS/HSD patients affecting their quality of life [111,116], though specific molecular pathways or mediators of pain are still unknown.

As further evidence of unbalanced inflammatory responses in patients' cells, transcriptome revealed a decreased expression of the *NR4A* nuclear receptors (*NR4A1*, *NR4A2*, *NR4A3*), which act as transcriptional regulators of inflammatory responses mediated by NF-kB signaling [159–161]. These receptors attenuate inflammatory events through inhibition of the NF-kB nuclear translocation and induction of the expression of its inhibitor NFKBIA, which, in turn, blocks the NF-kB nuclear localizing sequence [159]. The concomitant decreased mRNA levels of *NR4A1* and *NFKBIA* in patients' cells may be related to aberrant NF-kB signaling.

Despite additional work being required to support these hypotheses, our findings depict the complex sequence of transcriptional events that should stimulate more investigations to provide new insights into the pathomechanisms underlying the molecular networks related to aberrant inflammatory responses associated with hEDS/HSD.

#### **12. Emerging Aspects of hEDS**/**HSD Pathophysiology by microRNAs Profiling**

Transcriptome analysis may be a valuable strategy also to delineate distinct molecular signatures related to the differential expression of microRNAs (miRNAs). miRNAs are small non-coding RNA molecules ranging from 20–25 nucleotides in length that act mainly as negative regulators of gene expression by promoting the degradation of target mRNAs or repressing their translation [162,163]. Aberrant miRNA expression has been reported in several pathological conditions including cancer, musculoskeletal disorders, painful peripheral neuropathies, and fibromyalgia [164–166].

Our previous expression profiling of hEDS/HSD fibroblasts identified 19 dysregulated miRNAs [123]. Here, we report some examples that might offer interesting clues about the possible involvement of miRNAs in the regulation of potential targets and molecular pathways related to inflammation and Wnt signaling that seem to have a role in the disease pathogenesis.

The most overexpressed miRNA in patients' cells was the miR-378-3p, which is considered a modulator of the epithelial-to-mesenchymal transition and is associated with inflammation and fibrosis through the positive modulation of NF-kB and TNFα pathways [167,168]. miRNA-224, which is also up-regulated in hEDS/HSD cells, is associated with the activation of the Wnt/β-catenin signaling through the inhibition of the expression of glycogen synthase kinase 3β and SFRP2, which are known Wnt suppressors [169]. Therefore, it is likely reasonable to assume that the enhanced expression of this miRNA may contribute to the decreased mRNA level of *SFRP2* observed in hEDS/HSD cells, supporting the hypothesis that the aberrant signaling of the Wnt/β-catenin axis might play a role in the disease mechanisms of hEDS/HSD.

We previously demonstrated the involvement of the ILK in the fibroblast-to-myofibroblast switch of hEDS/HSD cells. The ILK acts downstream of the phosphatidylinositol 3-kinase signaling pathway and negatively regulates the action of the glycogen synthase kinase 3β by phosphorylation of a specific serine residue, further strengthening our assumption that Wnt/β-catenin signaling is involved in hEDS/HSD pathogenesis [49,50]. Consistently, patients' fibroblasts show a reduced expression of the miRNA-23a, which is implicated in the Wnt pathway regulation as well, by inhibiting the expression of FDZ5 and FDZ7 receptors [170]. Decreased levels of this miRNA were also found in cerebrospinal fluid and serum of patients with fibromyalgia [171], a painful disorder in differential diagnosis with hEDS/HSD [172], which might suggest a possible involvement of specific miRNA signatures or a common disease pattern in both conditions. In addition, the altered expression of this miRNA in synovial fibroblasts of psoriatic arthritis patients, results in the enhanced expression of pro-inflammatory mediators and matrix degrading enzymes, further promoting joint degeneration and synovial inflammation [173].

Several studies highlighted the contribution of miRNAs in the modulation of the expression of ECM structural proteins and related signaling molecules, thus emphasizing the close relationship between ECM homeostasis and inflammatory pain related conditions [174–176].

As the expression of miRNAs can be modulated to mediate the expression of their target genes, in-depth in vitro studies on a large cohort of patients' cells could provide further evidence on mechanisms of action of miRNAs and their impact on diverse target genes and altered pathways relevant for the pathophysiology of hEDS/HSD, thus offering new perspectives to identify potential molecular therapeutic targets.

#### **13. Conclusions and Perspectives**

Transcriptome and in vitro analyses on cEDS, vEDS, and hEDS/HSD dermal fibroblasts expanded the knowledge about molecular mechanisms involved in the pathophysiology of these connective tissue disorders (Figure 1).

**Figure 1.** Schematic illustration summarizing the processes likely involved in the pathogenesis of cEDS, vEDS, and hEDS/HSD derived from transcriptome and in vitro studies of patients' skin fibroblasts.

Our findings indicate that these cells share a deregulated expression of many matrix-related genes and a widespread disarray of several ECM structural constituents, thus highlighting the functional relevance of a proper organization and function of the ECM in providing stability to connective tissues. In cEDS and vEDS dermal fibroblasts, the pathological ECM turnover is directly caused by the underlying molecular defect causing abnormal expression of collagen V and collagen III, which, in turn, perturbs key physiological processes critical for collagen processing itself and to maintain cell homeostasis. In the absence of a known genetic etiology, the abnormal ECM organization present in hEDS/HSD cells may be a functional consequence of excessive remodeling due to increased levels of ECM-degrading enzymes and concomitant acquisition of a pro-inflammatory myofibroblast-like phenotype. hEDS/HSD transcriptome profiling for the first time has shed light on different pathobiological aspects of the disease. The dysregulated expression of genes involved in cell-matrix interactions and specific intracellular signaling pathways may have a role in the phenotypic switch of hEDS/HSD cells. Transcriptional changes of different genes and miRNAs involved in molecular pathways related to pain and inflammatory response might provide further clues to dissect the intricate biological events involved in chronic and musculoskeletal pain affecting hEDS/HSD patients. To deepen the knowledge on hEDS/HSD pathophysiology, proteome profiling of patients' cells is currently ongoing to decipher the complex protein network and identify potential bioactive molecules involved in the disease pathogenesis to offer therapeutic options for hEDS/HSD patients.

**Author Contributions:** Conceptualization: N.C., M.R., M.C.; writing and original draft preparation: N.C., M.R., M.C.; writing, review and editing: N.C., M.R., N.Z., M.C.; supervision: M.C.

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

**Acknowledgments:** The authors wish to thank the patients for their kind availability for these studies and the Fazzo Cusan family for its generous support.

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

#### **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*
