In 2019, the Nosology Committee of the International Skeletal Dysplasia Society published a new edition of the Nosology and Classification of Genetic Skeletal Disorders (Mortier et al., 2019). This 2019 version covers 461 different diseases divided into 42 groups according to their clinical, radiographic, and/or molecular phenotypes.

The application of massively parallel sequencing technology has led to the discovery in the last few years of a great number of genetic defects responsible for skeletal disorders. To date, the molecular bases have been identified for 425/461 (92%) of these disorders. In total, pathogenic variants affecting 437 different genes encoding enzymes, extracellular matrix (ECM) proteins, membrane transporters, cilia proteins, signal transduction proteins, and transcription factors have been found.

In this review, we will be focusing on the skeletal dysplasias caused by defects in the glycosaminoglycan (GAG) biosynthesis and, more specifically, on the group of chondrodysplasias with multiple dislocations (CMD), listed in groups 20, 4, and 25 in the International Classification on Genetic Skeletal Disorders (Mortier et al., 2019). They form a group of severe disorders characterized by joint laxity and multiple dislocations affecting large joints (such as hip, knee, and shoulder), severe short stature of pre- and post-natal onset, hand anomalies, and/or vertebral anomalies. Common radiographic features include advanced carpal and tarsal bone age and exaggerated trochanters giving a monkey wrench appearance of the femoral neck (Figure 1). Additional skeletal features, for instance, epiphyseal or metaphyseal changes, specific facial dysmorphisms, and cleft palate, are often part of the clinical presentation. A variable combination of other clinical features such as loose or old-appearing skin, congenital heart defects, teeth anomalies, intellectual disabilities, and obesity can also be observed in those patients. Up to now, more than 25 syndromes with autosomal recessive inheritance patterns have been described (Table 1). CMD has mostly been linked to pathogenic variants in genes implicated in the biosynthesis of proteoglycan (PG). PGs are large macromolecules that are widely expressed in multicellular organisms, present in the ECM or at the cell surface. They consist of a core protein and one or more covalently linked polysaccharides, called GAGs. GAGs are large linear polysaccharides composed of repeated disaccharide units consisting of amino sugar, either N-acetylglucosamine (GlcNAc) or N-acetylgalactosamine (GalNAc), and uronic acid, either glucuronic acid (GlcUA) or iduronic acid (IdoUA), except keratan sulfate (KS) in which disaccharide units consist of GlcNAc and galactose. Hyaluronan (HA) is a non-sulfated glycosaminoglycan and is not attached to any core protein. It is synthesized by a specific synthesis pathway taking place at the cell membrane. Sulfated GAGs are classified into four groups based on the composition of their disaccharide units: chondroitin sulfate (CS), dermatan sulfate (DS), KS, and heparan sulfate. In addition, these GAGs undergo further modifications, such as sulfation at various positions of the chain and epimerization of uronic acid (Lindahl et al., 2017).

PGs are highly diverse ECM components. Indeed, they can be composed of different core proteins with one or more GAG chain(s) of various subtypes and are subjected to variable degrees of post-translational modifications, including glycosylation and sulfation. Altogether, this leads PG to have a multitude of biological functions. Indeed, PGs promote ECM assembly by interacting with other ECM components, regulate ECM physical properties, and serve as a reservoir for various growth factors (Schaefer and Schaefer, 2010; Iozzo and Schaefer, 2015). In particular, PGs are highly expressed in cartilage ECM and play a major role in chondrocyte maturation and bone formation through endochondral ossification. PGs are also, through their ability to bind and retain water in the matrix, a critical component of articular cartilage, ensuring adequate mechanical properties and integrity maintain of articular cartilage (Martínez-Moreno et al., 2019).

The GAG biosynthesis is a complex process implicating the action of multiple enzymes (Figures 2, 3, Table 2) and that, although it is initiated in the endoplasmic reticulum (ER), occurs mainly in the Golgi apparatus cisternae (Prydz, 2015). GAG biosynthesis is initiated in the ER by the attachment of a xylose (Xyl) residue, using uridine diphosphate (UDP)-Xyl as a donor, to specific serine residues of the freshly synthetized PG core protein by β-xylosyltransferases encoded by XYLT1 or XYLT2 (Götting et al., 2007). After Xyl addition and shipment of the xylosylated protein into the Golgi apparatus, a linkage tetrasaccharide is formed by the transfer of two galactose (Gal) residues from UDP-Gal and one GlcUA from UDP-GlcUA via the sequential action of β1,4-galactosyltransferase-I (GalT-I), β1,3-galactosyltransferase-II (GalT-II), and β1,3-glucuronosyltransferase-I (GlcAT-I), encoded by B4GALT7, B3GALT3, and B3GAT3 (Okajima et al., 1999; Pedersen et al., 2000; Bai et al., 2001). Some modifications may occur on the linkage tetrasaccharide, including 2-O-phosphorylation of Xyl residue along with sulfation of the first Gal residue at the C-6 position and of the second Gal residue at the C-4 or C-6 position (Gulberti et al., 2005). The phosphorylation, which can be transient, is catalyzed by a GAG-Xyl kinase encoded by FAM20B (Koike et al., 2009). This phosphorylation together with sulfation may influence the catalytic activity of GalT-I, GalT-II, and GlcAT-I and thus the linkage region assembly and subsequent GAG elongation (Wen et al., 2014).

Once the linkage region is completed, two types of reactions occur and determine the type of GAG being synthetized: addition to the linkage tetrasaccharide of either a β4-linked GalNAc, which will initiate CS/DS assembly, or an α4-linked GlcNAc, which will initiate HS assembly (Lindahl et al., 2017).

In CS, the process starts with the transfer of a GalNAc from UDP-GalNAc to the last GlcUA residue of the linkage region by specific β1,4-N-acetlygalactosaminyltransferases, CSGalNAcT1 and CSGalNAcT2 (encoded by CSGALNACT1 and CSGALNACT2, respectively) (Uyama et al., 2002; Sato et al., 2011). CS chains are then polymerized by the action of one or more enzymes having both β3 glucuronosyltransferase and β4 N-acetylgalactosaminyltransferase activities, the chondroitin synthases (CHSY1-3) (Kitagawa et al., 2001). During this step, the chondroitin polymerizing factor, lacking independent activity, will interact with the chondroitin synthases and enhance the CS elongation (Izumikawa et al., 2008). CSs are then subjected to modifications such as epimerization and sulfation throughout the GAG synthesis process or just before PG secretion (Lindahl et al., 2017). CS sulfation in an elaborate process involved multiple sulfotransferases, three chondroitin-4-O-sulfotransferases (CHST11-13) for C4 sulfation of GalNac residues, two chondroitin-6-O-sulfation (CHST3 and 7) for C6 sulfation of GalNac residues, one GalNac-4-O-sulfate-6-O-sulfotransferase (CHST15) for sulfation of disulfated GalNAc, and one uronosyl-2-O-sulfotransferase for C2 sulfation of GlcUA (Mizumoto et al., 2013).

DSs are generated from CS by C5 epimerization of GlcUA to IdoA by two DS epimerases (DSE1–2) and sulfation in distinct positions by dermatan-4-O-sulfotransferase (CHST14) and uronosyl-2-O-sulfotransferase (Malmström and Aberg, 1982).

In HS, the assembly is initiated by the addition of a GlcNAc residue by an α1,4-N-acetylglucosaminyltransferase-I (GlcNAcT-I) encoded by EXTL3 (Kim et al., 2001). HS further elongation is carried out by HS polymerase complex formed by two enzymes with N-acetylglucosaminyltransferase and glucuronyltransferase activities and encoded by EXT1 and EXT2 (McCormick et al., 2000). HS then undergoes extensive modification reaction creating clusters of sulfated domains interspersed with an unsulfated region (Esko and Selleck, 2002). Those modifications are initiated by N-deacetylase/N-sulfotransferases (NDST1–4), which induce the N-sulfation of 40–50% of GlcNAc to N-sulfoglucosamine (GlcNS) followed by conversion of adjacent GlcA to idorunate (IdoA) by a glucuronyl epimerase. O-sulfotransferases can then modify these GlcNS/IdoA rich domains. HS2ST1 catalyzes the 2-O-sulfation of IdoA residues. IdoA(2S)-GlcNS can then be further modified by the addition of 6-O-sulfate and less frequently by addition of 3-0-sulfate groups to the GlcNS residues, by the action of 6-O-sulfotransferases (HS6ST1-3) and seven 3-O-sulfotransferases (HS3ST), respectively.

The sulfation of GAG is a crucial process during PG synthesis and is required for GAG physiological functions. In the Golgi apparatus, sulfotransferases use the 3′-phosphoadenosine 5′-phosphosulfate (PAPS) as a universal sulfate donor to transfer sulfate to specific residues of GAG chains (Paganini et al., 2020). PAPS is synthetized in the cytosol from adenosine triphosphate (ATP) and inorganic sulfate. The latter is transported from the extracellular environment into the cells through a sulfate/chloride antiporter named SLC26A2 (Hästbacka et al., 1994). PAPS synthesis takes place in two sequential steps by the action of a bifunctional enzyme, the PAPS synthase (Xu et al., 2000). The ATP sulfurylase activity firstly catalyzes the production of adenosine 5′-phosphosulfate (APS) from sulfate and ATP; subsequently, APS kinase activity produces PAPS from APS and ATP. Once PAPS is synthetized in the cytosol, it is translocated into the Golgi by two specific PAPS transporters (PAPS transporters 1 and 2) (Kamiyama et al., 2003, 2006). As a consequence of sulfotransferase activity, PAP is released and can inhibit those sulfotransferases via negative feedback. To prevent this, PAP is rapidly degraded into adenosine monophosphate and phosphate by a Golgi resident adenosine 3′, 5′-biphosphate 3′-phosphatase (gPAPP), encoded by IMPAD1 (also known as BPNT2) (Frederick et al., 2008).

Nucleotide sugars such as PAPS are synthetized in the cytoplasm and have to be transported into the Golgi by specific carriers, such as SLC35D1 or SLC35A3, that by an antiport mechanism will export nucleoside monophosphates in the cytosol (Muraoka et al., 2001; Maszczak-Seneczko et al., 2013). This nucleotide sugar import seems to be a rate-limiting step as increased UDP-N-acetylhexosamine availability leads to enhancement of the incorporation into glycoconjugates (Pels Rijcken et al., 1995).

GAG elongation and modification reactions probably colocalized within the Golgi cisternae, most likely by the formation of supramolecular complexes that coordinate these reactions. A correct conformation of Golgi cisternae and organization of their enzymatic content, as well as an adequate Golgi environment, i.e., a properly established pH gradient and concentration of ions such as Ca2+, are also required for correct GAG formation (Prydz, 2015). Any disturbances of this chain of reactions will lead to the incapacity of a cell to construct correct glycanic chains.

So far, up to 27 distinct human genetic disorders have been associated with pathogenic variants in 23 genes encoding proteins implicated in GAG biosynthesis (Table 1, Figures 2, 3). With few exceptions, such as EXT1/EXT2 or SLC35D1, the vast majority of the pathological variants identified in these genes are responsible for skeletal dysplasia associating short stature and joint laxity and/or large joint dislocations, characteristic of the CMD group (Table 1, gray rows). Clinical features of inborn errors of GAG biosynthesis, as well as the phenotype of existing related deficient animal models, are described in Table 1. The functional consequences of these inborn errors on GAG or PG synthesis, evidenced in patient samples, are listed in Table 2.

In the following section, we will focus on genes implicated in CMD.

CDG are an expanding group of rare, multisystem, underdiagnosed heterogeneous diseases caused by deficient or improper synthesis or attachment of glycans to proteins and lipids. More than 130 inherited disorders have been identified so far, most of them following an autosomal recessive inheritance pattern (Ondruskova et al., 2021).

Glycosylation is a very important post-translational modification, as it is estimated that at least 2% of the human genome codes for proteins involved in this vital biochemical pathway (Ng and Freeze, 2018). Due to the involvement of glycosylation in various cellular processes, CDG are characterized by multiorgan dysfunction and variable clinical phenotype. Therefore, CDG result in a broad spectrum of pathologies, including skin laxity, skeletal dysplasia, congenital heart defects, neurodevelopmental disorders, and endocrine abnormalities. The most common clinical manifestations include developmental delay, failure to thrive, microcephaly, coagulopathy, and abnormal brain magnetic resonance imaging, including cerebral and/or cerebellar atrophy, cell migration abnormalities, and immune dysfunction (Francisco et al., 2019).

CDG are classified into two groups, CDG I and CDG II. CDG I are defects in the glycan assembly and in the attachment of glycans to proteins in the ER. CDG II are defects in the processing of the already assembled glycans in the Golgi apparatus resulting in truncated and abnormal glycan structures. As it is now considered that glycosylation processes occurring in the ER are required for good folding and stability of the glycoprotein, whereas glycan remodeling in the Golgi will finely regulate the functionality of the protein, CDG I and CDG II will differently affect glycoprotein functions (Freeze, 2007).

Most protein glycosylation disorders are due to defects in the N-glycosylation pathways, but they can also be due to defective O-linked glycans (O-Mannose, O-Glucose, O-Fucose, O-GlcNac, and O-GalNac), GAG, glycosylphosphatidylinositol, and glycolipids (Freeze et al., 2014). As at the Golgi level, the different glycosylation processes coexist, alterations of the Golgi environment/organization, and inadequate supply in common subtrates, will lead to combinations of two or more affected glycosylation pathways. This is well-illustrated by CMD due to mutations in genes coding for transporters or Golgi proteins, i.e., SLC35A3, SLC10A7, and TMEM165, for which alterations in both N-glycosylation and GAG biosynthesis were detected (Foulquier et al., 2012; Edvardson et al., 2013; Bruneel et al., 2018; Dubail et al., 2018).

The group of CMD includes several complex syndromes with overlapping features and other clinical signs that are often non-specific, such as intellectual disability or cardiac defects. Even if few features are typical for some disorders (Table 1), characteristic hand anomalies in CANT1, IMPAD1, CHSY1, and TGDS-related conditions or amelogenesis imperfecta in dysplasia due to mutations in SLC10A7, for example, may help clinicians to orientate the clinical diagnosis. For the vast majority of CMD, phenotype–genotype correlations are still incomplete. Molecular or enzymatic assays to detect specific defects are thus crucial for final diagnostic. However, simple and affordable laboratory tests for use in screening are still missing, and, currently, clinicians must rely on research laboratories to perform the clinical molecular and biochemical tests to confirm the diagnosis. Many research teams have developed and used various techniques to evaluate quantitatively and qualitatively GAG synthesis. With these techniques, they were able to demonstrate in patient biological samples, for some PGs, a reduction of expression, concentration, or molecular weight or an alteration in the concentration of specific disaccharides, sulfated or unsulfated, constituting the GAG (Table 2). In most cases, analyses were performed on lysates or culture media from fibroblasts, cultured or not with radiolabeled substrates, after separation by high-performance liquid chromatography or gel electrophoresis (sodium dodecyl sulfate–polyacrylamide gel electrophoresis). Alternatively, altered migration of some low molecular weight PG was detected by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, and immunoblotting or abnormal GAG expression was revealed by immunohistochemistry or immunofluorescence techniques using primary antibodies against specific GAG residues. Being more easily obtained than fibroblasts, blood and urine samples are more convenient for biochemical testing to perform diagnosis in clinics. Only a few studies have studied the GAG in blood or urine from patients using high-performance liquid chromatography. Recently, abnormal electrophoretic migration profiles of bikunin, a serum PG with a single CS chain, were detected after immunoblotting in serum from patients with specific CMD and, more specifically, from patients with “linkeropathies,” indicating that bikunin is a potential biomarker, easily detectable, for these pathologies (Bruneel et al., 2018; Haouari et al., 2020). Moreover, for patients with mutations in SLC35A3, SLC10A7, or TMEM165, alterations in both N-glycosylation and GAG biosynthesis were detected (Foulquier et al., 2012; Edvardson et al., 2013; Ashikov et al., 2018; Dubail et al., 2018). This demonstrates that it could be useful to look for glycosylation defects in patients with CMD to help refine the diagnosis.

As for most of the skeletal and connective tissue disorders, treatments for CMD patients are still restricted to physiotherapy, orthopedic surgery, symptomatic treatment, or monitoring for potential complications to slow down or modify disease progression (Briggs et al., 2015; Marzin and Cormier-Daire, 2020). The development of new innovative therapies requires a strong comprehension of the molecular mechanisms leading to those disorders, mostly through extensive phenotypic analyses of in vitro and/or in vivo models. This is well exemplified by the use of N-acetylcysteine as a new therapeutic approach for patients with DTD. DTD is caused by pathogenic variants in a gene coding for SLC26A, a cell membrane sulfate-chloride antiporter, resulting in defective sulfate uptake leading to low cytosolic sulfate and subsequently PG undersulfation (Rossi et al., 1998). N-Ac, acting as an intracellular sulfate source for macromolecule sulfation, was thus tested, and preclinical studies in DTD mouse model showed promising results (Monti et al., 2015). N-Ac is currently being tested in DTD. Another example is given by TMEM165. Although the function of TMEM165 is still not completely understood, many observations suggest that TMEM165 has a role in Mn2+ homeostasis in the Golgi. As Mn2+ is a cofactor of glycosyltransferases, impaired Golgi Mn2+ homeostasis in the TMEM165-deficient patient is most likely responsible for the glycosylation defects. Strengthening this hypothesis, experiments performed in vitro have demonstrated that Mn2+ supplementation suppressed the glycosylation defects, and Mn2+ supplementation was proposed as potential new therapy for TMEM165-CDG patients (Dulary et al., 2017).

The recent evolution of genomic technologies has allowed a big step forward in the comprehension of pathophysiological mechanisms leading to rare genetic disorders, including skeletal disorders and the group formed by CMD. Studies of this latter group of pathology have proved a central role of GAG biosynthesis in the pathogenesis of these rare disorders. GAG biosynthesis has characteristic biochemical properties and affects many processes such as embryonic development and connective tissue formation and functions. It is also a complex and tightly regulated process that is still not yet completely clarified. Studies performed on patient samples, cell cultures, and animal models for human CMD have provided new insights on the GAG synthesis and on the physiologic functions of GAG in cartilage, bone, and connective tissues. However, further comprehensive approaches to the molecular pathogenesis involving GAG chains, in association or not with other glycosylation defects, are required to facilitate the development of new biomarkers for clinical screenings and innovative therapeutics for these diseases.

JD and VC-D have contributed to the writing of the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer AR declared a past co-authorship with one of the authors VC-D to the handling editor.