Morquio‐B disease: Clinical and genetic characteristics of a distinct GLB1‐related dysostosis multiplex

Abstract Background Morquio‐B disease (MBD) is a distinct GLB1‐related dysostosis multiplex involving the trabecular parts of long bones and spine, presenting a mild phenocopy of GALNS‐related Morquio‐A disease. Methods We analyzed 63 (n = 62 published) cases with MBD to describe their clinical, biochemical and genetic features. Results Forty‐one of 51 cases with informative clinical data had pure MBD including progressive growth impairment, kyphoscoliosis, coxa/genua valga, joint laxity, platyspondyly, odontoid hypoplasia. Ten of 51 had MBD plus neuronopathic manifestations including intellectual/developmental/speech delay, spasticity, ataxia dystonia. Corneal clouding, cardiac valve pathology, hepatosplenomegaly, spinal cord compression were infrequent and atlantooccipital dislocation, cardiomyopathy and cherry red spot were never reported. Urinary glycosaminoglycan and oligosaccharide excretion was consistently abnormal. Keratan sulphate‐derived oligosaccharides were only detected using LC‐MS/MS‐based methods. Residual β‐galactosidase activities measured against synthetic substrates were 0%‐17%. Among 28 GLB1 variants, W273 L (34/94 alleles) and T500A (11/94 alleles) occurred most frequently. W273L was invariably associated with pure MBD. Pure MBD also was reported in a case homozygous for R201H, and in the majority of cases carrying the T500A variant. Homozygous Y333C and G438E were associated with MBD plus neuronopathic manifestations. T82M, R201H, and H281Y, observed in seven alleles, previously have been found sensitive to experimental chaperones. Conclusion Data provide a basis for future systematic collection of clinical, biochemical, morphologic, and genetic data of this ultra‐rare condition.

1 | BACKGROUND GLB1-related disorders are caused by a deficiency of β-galactosidase, a lysosomal enzyme facilitating the degradation of complex carbohydrates bound to a variety of structurally unrelated molecules such as gangliosides, proteoglycans, and N-and Olinked glycoproteins. The spectrum of clinical phenotypes is wide: Type 1 (infantile) GM1-gangliosidosis (OMIM 230500) begins before age 1 year with hepatosplenomegaly, progressive loss of neurodevelopmental abilities and vision, cherry red macula spot, seizures, and dystonia/spasticity. Type 2 (late infantile/juvenile) GM1-gangliosidosis is characterized by a later onset of motor and cognitive regression. Type 3 (adult) GM1-gangliosidosis causes extrapyramidal signs, cardiomyopathy, and variable degrees of intellectual disability. 1 GLB1-related disorders are also associated with skeletal deformities (eg, kyphoscoliosis and short stature), which, like in other lysosomal storage diseases, nosologically have been classified as dysostosis multiplex. 2, 3 Morquio-B disease (MBD) (OMIM 253010) 4 is a distinct form of GLB1related disorder presenting with a specific type of dysostosis multiplex which has been known as Morquio syndrome since its first description by Morquio 5 and Brailsford. 6 Morquio syndrome is characterized by short stature with a disproportionally short trunk, kyphoscoliosis, pigeon chest (pectus carinatum), short neck, large appearing head with midface hypoplasia and mandibular protrusion, large appearing joints (elbows, wrists, knees, ankles), coxa and genua valga and flat feet. Joint laxity, corneal clouding, and cardiac valve disease and tracheal stenosis are additional findings. Characteristic radiological findings include platyspondyly and vertebral beaking, odontoid hypoplasia, spinal canal narrowing, hip dysplasia, dysplasia of the carpal and tarsal bones, as well as shortening and epi-and metaphyseal dysplasia of long bones (eg, shortening of the ulna and sloping of the distal ends of radius and ulna).
Currently two genetic conditions are known to cause Morquio syndrome: GALNS-related Morquio-A disease (OMIM 25300) and GLB1-related Morquio-B disease (MBD). Keratan sulfate is a proteoglycan that accumulates in both Morquio-A and Morquio-B disease.
The GLB1 gene contains 16 exons spanning more than 60 kb. The longest transcript variant (NM_000404.2) is a 2.5 kb mRNA giving rise to a 70 kDa precursor protein which is processed within the lysosomes into the 64-kD mature β-galactosidase enzyme protein. 4,7 GLB1 alternatively gives rise to a 2.0 kb mRNA transcript, formed by splicing out exons 3, 4, and 6 8 which encodes the elastin binding protein, a keyrecycling chaperone in the tropoelastin assembly process for elastogenesis in the extracellular matrix. 9 The β-galactosidase monomer consists of two β-domains and a TIM barrel domain, which together generate appropriate protein folding and functional integrity. 4,7,10 Mutations associated with Type 1/infantile onset GM1-gangliosidosis, for the most part, are located in the core protein region causing β-galactosidase instability, whereas mutations associated with milder phenotypes, such as types 2 and 3 GM1-gangliosidosis, tend to be on the protein surface. 7 Patterns and distribution of the accumulating substrates across the various tissues and organs are determined by the impact of the underlying GLB mutation on the molecular pathophysiology of the β-galactosidase protein. 4,11 While accumulation of GM1-gangliosides in the brain seems most responsible for neurologic manifestations in GM1-gangliosidosis, excretion of both skeletal and corneal forms of keratan sulfate has been shown in MBD and type 1 (infantile) GM1-gangliosidosis. 10 More than 150 pathogenic GLB1 variants are known with the vast majority being associated with GM1 gangliosidosis, 1 whereas the number of variants described in association with MBD is rather limited. W273L is the most frequent MBD allele. [10][11][12][13][14] Numerous other alleles have been found in both GM1-gangliosidosis and in MBD. 11,13,[15][16][17] but the type and degree of overlap between pure skeletal and neuronopathic phenotypes is hard to predict.
We performed a literature review to revisit all cases previously published as GLB1-related MBD, with the aims (a) to describe the clinical phenotype associated with MBD; (b) to compare the clinical data of all genotypes identified in this review against the classical W237L Morquio-B allele. Additionally, we reviewed published data on potentials of small molecules for allele specific rescue of β-galactosidase activity for those alleles identified in the reported MBD cases.

| METHODS AND RESULTS
Our literature review enrolled English language PubMedlisted publications and reports of cases diagnosed with MBD from its first description in 1976 18 to June 2018. Search terms included: "GM1-gangliosidosis," "Morquio-B," "Mucopolysaccharidosis," "Beta-galactosidase deficiency," "β-galactosidase deficiency," and "GLB1 deficiency." In Synopsis GLB1-related Morquio-B disease (MBD) is a distinct dysostosis multiplex resembling mild forms of GALNS-related Morquio-A disease and occurs as pure skeletal MBD and as MBD plus neuronopathic phenotype. The presence of at least one W273L allele determines pure MBD. Chaperone sensitivity has been shown in a variety of alleles associated with MBD. addition, we added data from three patients presented at the 13th International Symposium on Mucopolysaccharidoses and Related Diseases, Sauipe, Bahia, Brazil, August 13-17, 2014 19 and one thus far unpublished patient with MBD previously diagnosed at our center.
We identified 23 articles/reports containing information about 62 MBD cases. Including our own case (P1), 63 cases (22 male; 18 female; 23 gender not reported) met the inclusion criteria for our analysis: (a) clinical findings consistent with a Morquio phenotype; (b) diagnosis confirmed by demonstration of deficient β-galactosidase activity and/or a homozygous/compound heterozygous GLB1 mutation. We grouped all these cases according to availability of clinical information including skeletal and neuronopathic phenotype, as well as of GLB1 variant data (Table 1).
Fifty-one of 63 reported cases contained clinical data informing about the skeletal phenotype and presence or absence of a neuronopathic phenotype (Table 2). Forty-one of 51 cases had pure MBD presenting with skeletal features consistent with Morquio syndrome only, ten of 51 had neurologic/neurodevelopmental deficits in addition to typical skeletal features.
Most frequently reported skeletal features included short stature, kyphoscoliosis/platyspondyly, coxa valga, and genu valgum, odontoid hypoplasia and joint laxity/hyperextensible joints. In some cases, the skeletal features were described as progressive sponydloepiphyseal dysplasia. 18,32 Ulnar deviation of the wrist was reported in a 7-year-old girl 31 and a 7-year-old boy. 29 P1 had pronounced ligament instability in the ankles and wrists the latter resulting in a weak grip. The property of tooth enamel was reported in six cases: Beck et al 22 (n = 3), Groebe et al, 24 and van Gemund et al 32 and was normal in all cases but one (Guvec et al 25 ).
Spinal canal narrowing without myelocompression was reported in a 7-year-old male and a 10-year-old female, 18,31 and in a 40-year-old woman with myelocompression. 23 In a 15-year-old male with pure MBD 22 spinal malalignment had led to spastic paraplegia.
Growth parameters were reported in 21 of 51 patients ( Table 3). Short stature was a constant feature in adolescents and adults, whereas most of the younger patients had body heights within 1 SD of mean. The progressive nature of growth impairment is demonstrated by longitudinal growth data available from single patients.
Corneal clouding and cardiac valve pathology were reported in 20 of 51 cases ( Abnormal urinary excretion of glycosaminoglycans and of oligosaccharide containing glycoproteins was inconsistently reported. In P1 urinary oligosaccharides showed an abnormal band on thin layer chromatography and glycosaminoglycan excretion was mildly elevated (18 mg/mmol creatinine; reference range < 15) (Cetylpyridinium chloride test). Keratan sulfate was undetectable on urinary glycosaminoglycan electrophoresis, but Cases without informative clinical data (data on skeletal or neuronopathic or both phenotypes missing) 12 Hinek  keratan sulfate disaccharides were clearly elevated (28.5 μg/mg creatinine; reference range 0.24-2.71) upon UPLC/MS/MSbased determination. 36 Ultrastructural examination of a skin biopsy was reported in one single case 18 showing interstitial and cytoplasmatic U-shaped lamellar inclusions but absence of lysosomal inclusions found in GM1 gangliosidosis. 38 In Holzgreve et al 39 description of patients with Morquio syndrome, Adler-Reilly granular abnormalities were found in blood smears of patients with Morquio-A disease but not in the those with MBD.
Quantitative values of β-galactosidase activities were available in 49 of 51 cases with informative clinical data (β-galactosidase was measured but activity was not reported in the case published by Di Cesare et al 23 ; β-galactosidase was not measured in one of three siblings with MBD published by van Gemund et al 32 ). β-galactosidase activities were given in nmol/mg/min or in nmol/mg/min and/or as percent of normal. For reasons of comparability we calculated the percentage of the residual activities based on the respective mean of the normal range for all cases. Activities were measured either in white blood cells or in fibroblasts. Overall, residual β-galactosidase activities ranged between 0% and 17% in pure MBD cases and between 2% and 8.7% in MBD plus cases (Table 2).
In 47 of 63 cases genotypic information was available, harboring 28 different GLB1 variants. The most frequent alleles were W273L (34/94 alleles) and T500A (11/94 alleles). The characteristics of the 28 variants are shown in Table 4. Twenty-five of 28 variants were missense, two were frameshift, and one was splice site. Seven variants found in the reviewed cases had previously been tested for chaperone sensitivity with three of them being reported chaperone sensitive (T82M, R201H, H281Y).
Information about the clinical (skeletal and neuronopathic) phenotype in conjunction with the underlying GLB1 mutations was available in 38 cases (Table 2). Twenty-nine of 38 had pure MBD (dysostosis multiplex type Morquio syndrome without evidence of neuropathic involvement). Additional neuronopathic manifestations (MBD plus) were reported 10 of 51 cases (nine of 38 with known genotype and one of 13 with unknown genotype) ( Table 2). Neuronopathic manifestations included: developmental delay/intellectual disability, loss of motor and cognitive skills with onset in late infancy, delayed/ impaired speech, ataxia, spasticity, dystonia, myoclonia, choreoatetosis. Brain MRI was reported in only one of the MBD plus cases (Roze et al, 17 case 2) and was normal. These 38 cases harbored 21 of the 28 GLB1 variants identified in this review.
Four of 21 variants were present in homozygosity: W273L (12 cases/five families); R201H (one case/one family); G438E        and inability to walk. He had an increased tone of the limbs and brisk tendon reflexes. CV 3 (male) was diagnosed with MBD based on characteristic skeletal abnormalities at the age of 4. While at the age of diagnosis his motor and cognitive development was normal, he gradually lost his ability to speak and he became tetraspastic after the age of 10 years. CV 4a, 4b, 4c (T500A/c.1577dupG = p.G526GfsX5) 19 demonstrate the phenotypic variability of the same genotype. Among three male unrelated patients with the same genotype, two (ages 7 and 39) had skeletal features consistent with MBD without neuronopathic involvement, while one (age 31) had psychomotor delay and speech difficulties at age 3 and thereafter developed borderline intelligence and neurological regression. ß-galactosidase activities were not discriminative (4% in both cases with pure MBD and 2% in the MBD plus case).
CV 5 (R201H/S149F) (case 21 in Hofer et al 11 ) features an unspecific (non-Morquio) type of dystostosis multiplex, associated with mild intellectual disability. This 14-year-old male was described as "atypical" MBD. He presented with mild intellectual disability and dorsolumbar kyphoscoliosis, but characteristic Morquio features (short, disproportionate stature, dysplasia of the odontoid, hip dysplasia, genua valga) were not present when assessed at 19 years. Corneal clouding, cherry red spots, cardiac involvement, and organomegaly were absent. CV 6 (R201H/R201H) (MBD2 Santamaria et al 14 ) features the sole described patient with pure MBD who is homozygous for a variant that previously has been described in association with neuronopathic (types 2 and 3) GM1 gangliosidosis. At the age of 16, this male patient exhibited skeletal features characteristic of MBD. He had a history of a normal development, his cognition was within normal range and neurologic signs and symptoms suggestive of a primary neuronopathic course were absent.

| MBD phenotypes
The results of this study have reinforced the general understanding that MBD is a distinct variant of GLB1-related disease with an axial and appendicular dysostosis multiplex as initially described in GALNS-related Morquio-A disease. However, while Morquio-A disease is invariably associated with normal intellectual development and lifelong absence of primary neuropathic manifestations, only just below 80% of the GLB1-related MBD cases presented with a pure skeletal phenotype. The remaining 10 cases showed additional primary neuronopathic manifestations. These findings indicate that GLB1-related MBD occurs in two forms: pure MBD and MBD plus neuronopathic manifestations.
Results also have reinforced the general understanding that the skeletal manifestations are mild in MBD when compared to typical Morquio-A disease and clinically indistinguishable from mild Morquio-A variants. 39,42,43 Along with this notion, we previously could show that the height of adult MBD patients is significantly less compromised than of those with typical Morquio-A disease. 44 Notably, three of the five adults (24-39 years old) depicted in Table 3 had only a mildly impaired body height (minus 1.2 SD-minus 2 SD). Conversely, among the six individuals younger than 6 years, only the patient reported by Sheth et al 29 had early onset dwarfism with a body height at minus 3.8 SD at the age of 3 years.
In cases with mild skeletal involvement, such as CV 5 who was described as "atypical Morquio-B" a distinction between MBD and GM1-gangliosidosis with unspecific dysostosis may be challenging. Therefore, for the sake of a precise classification, we recommend that the diagnosis of MBD should be assigned only if there are ≥3 radiological findings characteristic of Morquio syndrome, such as platyspondyly and vertebral beaking involving all segments of the spine, odontoid hypoplasia, epi-and metaphyseal dysplasia of long bones, and hip dysplasia.
Other manifestations, which are typically observed in GALNS-related Morquio-A disease [45][46][47] were not reported at all (such as atlantoaxial subluxation, hearing impairment, tracheal stenosis) or were reported only in a minority of cases (such as spinal cord narrowing, myelocompression, hepatosplenomegaly, cardiac valve pathology, tooth enamel abnormalities). Cardiomyopathy and retinal cherry red spot, which typically occur in GM1-gangliosidosis, 1 were not reported in any of the MBD cases. Corneal clouding was observed mainly in cases who had mutations in the catalytic domain essential for keratan sulfate substrate processing such as W273L and Y333C. 4,10,33,48 Impaired elastogenesis has been shown in GM1 gangliosidosis and MBD. 33,49,50 Apart from a single study in a skin biopsy of a patient with MBD, 18 biochemical and morphologic studies on extracellular matrix or bone pathology have not been performed for MBD. Future studies similar to those performed for numerous other LSDs 51 and for Morquio-A disease 52,53 may serve for a better understanding of the clinical differences between Morquio-A disease and MBD (eg, why surgical operations in MBD patients differ significantly from those with Morquio-A disease both in regards to the types of intervention, and in regards to the age at which these surgeries become necessary. 44 Neuronopathic manifestations in MBD span from an early onset global developmental delay with delayed achievement of motor milestones, speech delay to intellectual disability, progressive spasticity, and dystonia (CV 1). Onset of neurocognitive deficiencies can be later in life (CV 2), with skeletal findings being the first red flag (CV 3). Clinicians should bear in mind that patients diagnosed with pure MBD in early childhood might develop neuronopathic problems later on.

| MBD genotypes
W273 L was invariably associated with pure MBD. The amino acid residue Trp-273 resides at the entrance of the ligand-binding pocket of β-galactosidase, which acts as a holder of substrates for catalytic reaction. W273L affects the degradation of keratan sulfate more severely than the turnover of GM1-ganglioside, explaining the predominance of skeletal manifestations. 7,[10][11][12]34,54,55 Pure MBD also was reported in single a case homozygous for R201H (CV 6). Arg-201 is located on the lateral face of the TIM barrel domain, which is far from the ligandbinding pocket 7,56 thus not specifically affecting the catalytic activity towards keratan sulfate. It has rather been suggested that the R201H mutation results in a mislocalized, unstable precursor protein. 11,41 Several cases were found where the R201H allele was associated with type 2 GM1-gangliosidosis, 1 while its association with pure MBD 12 remains to be confirmed in more cases.
The other GLB1 variants found in homozygosity (G438E and Y333C), were associated with MBD plus ( Table 2). G438E causes an abnormal complex formation alone or coupled with keratan sulfate binding 21 with a relatively high (6.1%) residual activity. 11 Results of enzyme activity assays using different substrates suggest that Y333, similar to W273L, affects the active site of β-galactosidase rather than affecting the enzyme stability 16 comparable to D332, the adjacent amino acid residue, which is directly involved in the catalytic reaction. 11 Y333H is not invariably associated with MBD, as homozygous cases have been described with Type 2 GM1-gangliosidosis lacking the specific features of Morquio syndrome. 11 After W273L, T500A was the second frequent allele occurring in heterozygosity in 11/58 alleles. In the 38 cases with clinical and genetic information, six of eight compound heterozygous cases presented with pure MBD.

| β-galactosidase activity and biomarkers
We were not able to establish a correlation between residual β-galactosidase activities, genotypes and phenotypes. The main reason for the inability to discriminate molecular characteristics of the various GLB1 mutations is the use of synthetic substrates (eg, 4-MU-β-galactoside) for the determination of β-galactosidase activity, which only allows a rough discrimination between zero residual activities (eg, in infantile GM1-gangliosidosis), and activities up to 2%-10% (eg, in late onset GM1-gangliosidosis and MBD). 11 To precisely determine the biochemical characteristics of β-galactosidase mutants, measurements using natural substrates are needed. However, such measurements are laborious and have rarely been performed. 54,55 Technical variations in the enzyme assays across the various labs and the type (white blood cells, fibroblasts) and quality of samples used also contribute to the poor correlation of β-galactosidase activity with the genotype.
Likewise, limited information is has been found regarding a correlation between chemical biomarkers and the genotype. Keratan sulfate is the main storage product in MBD, however analytical challenges imposed by the use of traditional methods may explain why in the cases reviewed here keratan sulfate was either not determined or information was mostly restricted to its presence or absence. Quantitative measurements of keratan sulfate using LC-MS/MS-based technologies have only recently become available. As shown in P1, urinary keratan sulfate accumulation could only be shown upon LC-MS/MS-based analysis but not upon traditional glycosaminoglycan electrophoresis. Other studies employing LC-MS/MS-based technology have shown an accumulation of mono-and disulfated keratan sulfate species in blood and urine of single MBD patients 57 and a correlation with clinical severity has been shown in Morquio-A patients. 45 Further studies are needed to determine age and phenotype related biomarker profiles in MBD patients.

| Chaperone sensitivity
Several pharmacological chaperones acting on β-galactosidase including galactose, N-octyl-4-epi-beta-valienamine (NOEV), alkylated or fluorinated derivates of desoxynojirimycine (DGJ), and (5aR)-5a-C-Pentyl-4-epiisofagomine have been tested against numerous GLB1 mutant enzymes. [58][59][60][61][62] As a general rule, chaperone responsive mutant proteins harbor intact catalytic sites but fail in achieving full maturation or appropriate localization in the lysosomes due to protein misfolding or lack of protection by protective protein/cathepsin A. 37,40,41,63 Three of the GLB1 alleles identified in this review (T82M, R201H, H281Y) have been shown in the literature to be chaperone responsive. The most pronounced response was observed in the R201H allele using DGJ derivatives as chemical chaperones. Human fibroblasts carrying this variant in homo-or heterozygosity showed an up to 12.5-fold increase of basal β-galactosidase activity resulting in 30% of normal control activity. 40,41 According to theoretical considerations 64 and evidence shown in cell cultures, 63,65 residual enzyme activities beyond 10% to 15% may be sufficient to avoid substrate accumulation. A comparable magnitude of β-galactosidase enhancement has been reported by 59 in 10 out of 15 GLB1-deficient fibroblast lines tested against a 4-epi-isofagomine derivative. Interestingly, six fibroblast lines carried at least one mutation at the amino acid residue Arg-201. Despite significant achievements in preclinical research, with the exception of the aminosugar Miglustat, 66 chaperone therapy yet has not been established for patients with GLB1-related conditions. W273L, the most frequent MBD allele, is not sensitive to chaperon rescue as it encodes for a catalytic mutant within an otherwise stable, normally trafficked and localized protein 12,40,67 (Paschke unpublished). Therefore only compound heterozygous individuals harboring a second chaperone-sensitive allele will benefit from this form of therapy. Substantial progress in the development of gene therapies for GLB1-related conditions 68 will benefit patients with variants not amenable to chaperone therapies.

| CONCLUSION AND OUTLOOK
Overall, this review of published cases with MBD has shown that MBD occurs as a spectrum of distinct skeletal and non-skeletal (neuronopathic) manifestations.
While there is a clear association between pure MBD and the W273L allele, further studies are needed to better determine genotype-phenotype correlations of MBD plus alleles as well as their role in elastogenesis and bone pathology.
Careful clinical phenotyping of this ultra-rare condition is important for elucidation of the natural history of MBD informing the choice of outcomes in future clinical trials. Clinical assessments should include a full skeletal survey with additional attention to the cranio-cervical junction, as well as a full clinical, neurologic, and neurocognitive exams, including a brain MRI. Biochemical phenotyping should include the determination of β-galactosidase activity in white blood cells or fibroblasts, as well as quantitative (LC-MS/MS based) determination of urinary glycosaminoglycans and keratan sulfate-derived oligosaccharides. We have started collecting data via an international patient registry 8 and are currently initiating repositories for longitudinal data and biological sample collection.

ACKNOWLEDGMENTS
The project was funded by The Priest Family Fund for Morquio-B research, a UBC-based stewardship grant. I.S.A. was supported by a fellowship grant from Umm Alqura University, Makkah, Saudi Arabia. We thank the Laboratory for Diagnosis of Lysosomal Storage Diseases (E. Paschke, W. Windischhofer, K Paul) for performing mutation analysis in P1.

CONFLICT OF INTEREST
S.S.-I. holds The Priest Family Fund for Morquio-B Research, a UBC-based stewardship grant. She has received educational grants from Biomarin, Shire, Recordati and she serves/served as PI in clinical trials and postmarketing registries sponsored by Actelion, Biomarin, Shire, Ultragenyx. I.A., N.Y., and E.P. have no conflicts to declare.

AUTHOR CONTRIBUTIONS
I.SA. performed the literature review, extracted and analyzed data, and wrote the manuscript. N.Y. coordinated Morquio-B related research, participated in manuscript writing and editing. E.P. contributed and critically reviewed data and biochemical/genetic data. S.S.-I. initiated this research project, and supervised progress of work and data analysis, analyzed data, contributed to manuscript writing, and edited the final version of the manuscript.

ETHICS APPROVAL
Not applicable.