Clin Genet 2010 © 2010 John Wiley & Sons A/S Printed in Singapore. All rights reserved CLINICAL GENETICS doi: 10.1111/j.1399-0004.2010.01379.x Short Report Phenotype determining alleles in GM1 gangliosidosis patients bearing novel GLB1 mutations

a a a Hofer D, Paul K, Fantur K, Beck M, Rouberge A, Vellodi A, Poorthuis BJ, DHofer , K Paul , K Fantur , b c Michelakakis H, Plecko B, Paschke E. Phenotype determining alleles in M Beck , A Rouberge , GM1 gangliosidosis patients bearing novel GLB1 mutations. A Vellodid, BJ Poorthuise, Clin Genet 2010. © John Wiley & Sons A/S, 2010 H Michelakakisf, B Pleckoa and E Paschkea GM1 gangliosidosis manifests with progressive psychomotor deterioration aDepartment of Paediatrics, Medical and dysostosis of infantile, juvenile, or adult onset, caused by alterations b in the structural coding for lysosomal acid ß-galactosidase (GLB1 ). University of Graz, Austria, Childrens’s Hospital, University of Mainz, Lysosomal In addition, allelic variants of this gene can result in Morquio B disease Disorder Working Group, Germany, (MBD), a phenotype with dysostosis multiplex and entire lack of cDepartment of Child Neurology, neurologic involvement. More than 100 sequence alterations in the GLB1 Trousseau Hospital AP-HP, Paris, gene have been identified so far, but only few could be proven to be France, dDepartment of Metabolic predictive for one of the GM1 gangliosidosis subtypes or MBD. We Medicine, Great Ormond Street Hospital for Children NHS Trust, London, UK, performed genotype analyses in 16 GM1 gangliosidosis patients of all e phenotypes and detected 28 different genetic lesions. Among these, Department of Medical Biochemistry, Academic Medical Centre, University of p.I55FfsX16, p.W65X, p.F107L, p.H112P, p.C127Y, p.W161X, p.I181K, Amsterdam, Amsterdam, Netherlands, p.C230R, p.W273X, p.R299VfsX5, p.A301V, p.F357L, p.K359KfsX23, and fDivision of Enzymology & Cellular p.L389P, p.D448V, p.D448GfsX8, and the intronic mutation IVS6-8A>G Function, Institute of Child Health, Agia have not been published so far. Due to their occurrence in homozygous Sophia Children’s Hospital, patients, four mutations could be correlated to a distinct GM1 Athens, Greece gangliosidosis phenotype. Furthermore, the missense mutations from Key words: GLB1 gene – GM1 heteroallelic patients and three artificial nonsense mutations were gangliosidosis – mutation analysis – characterized by overexpression in COS-1 cells, and the subcellular phenotype – genotype correlation localization of the mutant in fibroblasts was assessed. The Corresponding author: Dr Eduard phenotype specificity of 10 alleles can be proposed on the basis of our Paschke, Department of Paediatrics, results and previous data. Medical University of Graz, Auenbruggerplatz 30, A-8036 Graz, Austria. Tel.: +43 316 385 4035; fax: +43 316 385 4024; e-mail: [email protected]

Received 16 October 2009, revised and accepted for publication 11 January 2010

Two lysosomal storage diseases, GM1 gangliosi- complex carbohydrates in the lysosomal compart- dosis (GM1; MIM# 230500) and Morquio B ment. disease (MBD; MIM# 253010), are caused by GM1 is a clinically heterogeneous, neuroso- sequence alterations in a single gene, GLB1.They matic disease. In its infantile form (type I, OMIM result in functional deficits of acid ß-galactosidase #230500), early developmental arrest is observed (ß-Gal; EC 3.1.2.23; MIM# 611458), an followed by progressive neurologic deterioration cleaving terminal ß-linked residues from and spasticity (1). Macular cherry-red spots, facial

1 Hofer et al. dysmorphism, hepatosplenomegaly and general- the proposed intra- and intermolecular mecha- ized dysostosis are usually present. Juvenile (type nisms in pathogenesis. Recently, a structural model II, OMIM #230600) and chronic phenotypes (type of human ß-Gal was constructed using cristal- III, OMIM #230650) display a progressive neuro- lographic data for B.thetaiotaomicron ß-Gal as logic disease in childhood or early adulthood with a template (12). Based on an amino acid iden- minor or absent dysmorphic changes (1). On the tity with human ß-Gal of 36%, good correla- contrary, typical MBD patients completely retain tion between structural changes caused by amino their neurological functions, but develop severe acid substitutions in the ß-galactosidase molecule skeletal dysplasia, not distinguishable from attenu- and biochemical and clinical phenotypes was pro- ated manifestations of IV. posed for three mutations found in an infan- Impaired degradation of glycopeptides and gly- tile (p.G123R), a juvenile (p.R201C) and an cosaminoglycans including keratan sulfate in con- adult (p.I51T) GM1 patient. However, a rare nective tissues and visceral organs is supposed to allele (p.Y83H) was chosen for MBD, found cause growth retardation, corneal clouding, and in a single, heteroallelic MBD patient (genotype valve thickening in the heart as well as increased p.Y83H/p.R482C). Therefore, this model has to be urinary excretion of keratan sulfate. Although completed with data for the major common MBD these disease phenotypes were first described more allele, p.W273L (13) before it can be generally than 30 years ago, their pathogenesis is still not applied. fully resolved, most likely due to rather complex Nevertheless, for few of the over 100 detected molecular mechanisms involved. mutations of the GLB1 gene, distinct functional The GLB1 gene, located on deficits of its gene products can be suggested 3p21.33, is composed of 16 exons (2,3). Upon due to their occurrence in homoallelic cases: expression, results in two p.W273L, the common allele in MBD, has repeat- different transcripts, a 2.5-kb RNA coding for edly been found in homoallelic patients and enzymatically active acid ß-Gal and a minor determines the phenotype if present in heteroal- 2.0-kb RNA encoding the binding lelic patients (13,14). It is located in the cat- (EBP) (4), located in the cellular membrane and alytical domain and supposed to code for an involved in the assembly of tropoelastin into grow- enzyme with increased activity (>10% of nor- ing elastin fibers (5,6). It lacks exons 3, 4 and 6 mal) against synthetic substrates, strongly reduced and contains a unique stretch of 32 amino acids due to a different reading frame in exon 5. On the affinity toward keratan sulfate and normal degra- level of gene products, they can include effects dation of GM1 ganglioside (15). The enzyme pre- on both ß-Gal and EBP (amino acids 1–82 and cursor undergoes normal processing and trans- 245–678) or on ß-Gal alone (amino acids 83–153 port to the lysosomes (16,17). Some other com- and 185–245). mon alleles (p.I51T, p.T82M, p.R201H, p.R201C) Both GLB1 transcripts have been shown to form found in homozygous patients with juvenile or multiprotein complexes with neuraminidase, pro- adult GM1 also exhibit increased residual activi- tective protein/ (PPCA), and N-acetyl- ties against synthetic substrates. They were shown galactosamine-6-sulfate sulfatase either on the to express enzyme precursors with impaired sta- surface or in the lysosome, respectively (7,8). bility, diminished aggregation with other com- Mutations affecting residues on the surface of the ponents of the multienzyme complex and rapid ß-Gal molecule are therefore supposed to alter the subsequent degradation (18,19), whereas in severe interaction of ß-Gal or EBP with the other compo- infantile patients fibroblasts, ß-Gal activity against nents of the multiprotein complex. Several alleles synthetic substrates is very low (<1%) and a com- were shown to influence the multienzyme assem- plete lack of ß-Gal protein may be present (15). bly and aggregation (9–11). Moreover, some GM1 mutations were shown to ß-Gal cleaves terminal ß-linked galactose re- protect patients from developing the most severe sidues from structurally diverse natural substrates, form of GM1 (19–21). such as GM1 ganglioside, N- and O-linked Our group recently published further clinical, oligosaccharides and keratan sulfate in the lysoso- biochemical, immunohistochemical and molecular mal compartment. Mutations affecting the catalyt- data on GM1 and MBD patients and an expression ical site of the enzyme may differentially modify analysis of novel pathogenic alleles in cultured the degradation of natural substrates. cells (22). We now report on the detection of The tertiary structures of ß-Gal or EBP have, novel pathogenic alleles in the GLB1 gene and however, not been resolved. Therefore, single estimate the phenotype specificity of alleles in 10 mutations cannot certainly be correlated with heteroallelic patients, based on empirical data from

2 Phenotype determining alleles in GM1 gangliosidosis patients GM1 patients, expression studies and mutation type.

Materials and methods Patients After diagnosis of GM1 gangliosidosis in 16 1British 1German 5French 7German 0Turkish < < 3.5 Moroccan 0.4 Turkish 1.4 Greek 0.8 Turkish 3.7 Honduras 2.6 British n.a. German n.a. German

patients, genetic analysis of the GLB1 gene was 5–15 Pakistani performed. Their clinical data were collected by several coauthors (MB; AR, AV, BJP, HM, BP)

and by the colleagues mentioned in the acknowl- ot available. + edgements section using the same common ques- is study tionnaire (Table 1).

DNA isolation, amplification, and restriction analysis DNA was isolated from skin fibroblasts or blood using the QIAamp Blood Mini Kit (Qiagen, − −−+ n.a. n.a. n.a. Hilden, Germany). The genomic DNA was sub- n.a. n.a. n.a. n.a. 0.8 Pakistani jected to PCR for amplification of the 16 exons and + + + − −−+ + − −−+ + − −−+ + − +−+ + − −−+ + − −−+ + − −−− + − −−+ + + −−+ + − −++ + + −−+ the flanking regions of the GLB1 gene. PCR condi- + + −++ tions have recently been described (22). cDNA ref- erence sequence is GenBank NM 000404.2. The identified mutations were confirmed by restriction analysis (Table 2) with purchased from New England Biolabs (Frankfurt, Germany) and Roche (Vienna, Austria). For the exclusion of poly- morphisms, at least 100 control alleles were ana- lyzed for each novel mutation. The novel intronic mutation c.734-8A>G could not be found in 230 normal DNA samples.

Site-directed mutagenesis Introduction of mutations into the wildtype DNA was performed using the QuikChange II Site Directed Mutagenesis Kit (Stratagene, La Jolla, California) following the manufacturer’s manual. Template was the wildtype ß-Gal cDNA in the modified pcDX vector, kindly provided by A. Mor-

rone, Department of Paediatrics, University of Flo- Clinical phenotype Onset Age at diagnosis CNS CRS CI HS SA ß-Gal activity (%) Ethnic origin rence, Children’s Hospital ‘A.Meyer’, Italy, and A. d’Azzo, Department of Genetics and Tumor A Infantile GM1 6 months 9 months

Cell Biology, St. Jude Children’s Research Hos- > G Infantile GM1 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. British > pital, Memphis, TN. Mutagenesis was confirmed 1G +

by restriction analysis (Table 2). In addition to the a expression of the newly identified mutations, three

different artificial nonsense mutations in exon 16 A/c.245 (p.W582X, p.E620X, p.K659X) were constructed > p.K359KfsX23/ R442Q Adult GM1 4 years 16 years 1G + to assess residual enzyme activity of truncated ß- + Gal mutants.

Expression analysis Confluent COS-1 cells were transfected in 12-well 6 p.W273X/p.D448V Infantile GM1 n.a. 1 years 7 p.C127Y/p.C127Y Juvenile GM1 1 year 2 years 9 p.I55FfsX16/p.R201H Juvenile GM1 4 years 6 months 10 years 45 p.W161X/p.R208C p.C230R/p.C230R Infantile GM1 Infantile GM1 11 months 0 n.a. 2 months 8 p.F107L/p.D448GfsX8 Juvenile GM1 2 years 12 years + 2 p.R148S/c.734–8A 3 p.I181K/p.I181K Infantile GM1 6 months 1 years 1c.245 Sequence variations on DNA level are reported in Table 2. 1011 p.A301V/p.L389P p.R49H/p.G526GfsX5 Juvenile Adult GM1 GM1 1 year 6 months 2 2 years years 2 months 4 years 1213 p.K73E/p.R148C p.T82M/p.H112P Adult GM1 Adult GM1 16 years 6 years 16 years 15 years 141516 p.F357L p.W65X/p.L155R p.L155R/p.R299VfsX5 Adult GM1 Adult GM1 4 years 4 years 6 years 9 years trays with 1.2 μg wildtype or mutant DNA using Table 1. Clinical findings, genotype, residual ß-GalPatient activity measured in leukocytes or fibroblast cells, and ethnic origin of patients included in th Genotype CNS, central nervous systema involvement; CRS, cherry-red spots; CI, cardiac involvement; HS, hepatosplenomegaly; SA, skeletal affection; n.a., n

3 Hofer et al.

Table 2. Sequence alterations in the GLB1 gene of GM1 gangliosidosis patients

Coding effect Nucleotide change Exon Restriction site alteration References p.R49H c.146G>A2 −CfoI(23) p.I55FfsX16 c.163delA 2 +AluI This paper p.W65X c.195G>A2 −BsrI This paper p.K73E c.217A>G2 −EarI(23) p.T82M c.245C>T2 +MsII (21) IVS2+1G>A c.245+1G>AIVS2+HpaII (24) p.F107L c.319T>C3 −SspI This paper p.H112P c.335A>C3 +BanI This paper p.C127Y c.380G>A3 −BsgI This paper p.R148C c.442C>T4 −BsrBI (23) p.R148S c.442C>A4 +DdeI(25) p.L155R c.464T>G5 +NciI (26) p.W161X c.483G>A5 −MfeI This paper p.I181K c.542T>A5 +MseI This paper p.R201H c.602G>A6 −BsgI(20) p.R208C c.622C>T6 −BsmI(27) p.C230R c.688T>C6 +MaeII This paper IVS6-8A>G c.734-8A>GIVS6 −Tsp509I This paper p.W273X c.818G>A8 +AluI This paper p.R299VfsX5 c.895delC 8 −BglI This paper p.A301V c.902C>T8 +HphI This paper p.F357L c.1071T>G11 +HphI This paper p.K359KfsX23 c.1077delA 11 +HphI This paper p.L389P c.1166T>C12 +MspI This paper p.R442Q c.1325G>A13 −TaqI(28) p.D448V c.1343A>T13 −FokI This paper p.D448GfsX8 c.1342dupG 13 −SfcI This paper p.G526GfsX5 c.1577 1578insG 15 −AluI(29)

4 μl Lipofectamine 2000 (Invitrogen, Karlsruhe, Western blot analysis Germany), according to the manual. Negative con- For Western blot analysis of the mutant trol was the pcDX vector without insert. After 20 h ß-Gal proteins, 20 μg of total protein of COS-1 of incubation, the cells were harvested by scraping cell homogenate were separated with 10% SDS- and resuspended in 0.9% NaCl containing 0.01% polyacrylamide gel electrophoresis and subse- Triton X-100. The cell suspension was sonicated quently transferred to Invitrolon PVDF membranes for lysis and ß-Gal activity, and protein content (Invitrogen, Karlsruhe, Germany). Hybridization was determined in the supernatant. Fibroblasts as with α85 antihuman ß-galactosidase antibody well as COS-1 cells were maintained in Minimal directed against the precursor and the mature form Essential Medium (MEM) with Earle’s Salts (PAA, of ß-Gal (31), kindly provided by A. d’Azzo, Pasching, Austria) containing 10% fetal bovine Memphis, TN, and with a secondary phosphatase- serum, 400 μM L-glutamine, and 50 μg/ml gen- conjugated goat anti-rabbit lgG (whole molecule) ◦ antibody (Sigma–Aldrich, Taufkirchen, Germany) tamicine at 37 C and 5% CO2. ß-Gal and ß-hexosaminidase activity were was performed. The alkaline phosphatase color measured using the synthetic substrates reaction (AP Conjugate Substrate kit; Bio-Rad, Hercules, California) was applied to visualize the 4-methylum-belliferyl-ß-D-galactopyranoside and ß-Gal corresponding protein bands. BenchMark 4-methylumbelliferyl-N-acetyl-ß-D-glucosaminide prestained protein ladder was used to estimate (Sigma–Aldrich, Taufkirchen, Germany), respec- the protein size (Invitrogen, Karlsruhe, Germany). tively. Enzymatic hydrolase activity was measured Fibroblast cells Western blotting was performed with a luminescence spectrophotometer (Perkin using the same procedure, but 30 μg of total pro- Elmer, Wiesbaden, Germany) according to stan- tein of cell homogenate per gel lane were loaded. dard protocols. Determination of total protein con- For PPCA detection, a polyclonal PPCA antibody tent of the harvested cells was performed as (Abcam, Cambridge, UK) was used. Access to described (30). fibroblast cultures was limited to patients 3, 4, 5, 8,

4 Phenotype determining alleles in GM1 gangliosidosis patients 10 and 11. Therefore, Western blotting could only of the mutation site. Thus, they are all likely be performed in homogenates of these cell lines. to result in truncated, enzymatically inactive products. Upon recognition of the premature Immunostaining stop codon, the nonsense-mediated mRNA decay will also play an important role in the removal FITC-staining of the ß-Gal in healthy fibrob- of the incomplete RNA prior to of last cells and six GM1 gangliosidosis cell lines the truncated protein. Twenty-five percent of all (patients 3, 4, 5, 8, 10 and 14) was performed using detected genetic lesions occur in codons presenting the α85 antihuman ß-galactosidase antibody and a the potential CpG DNA methylation site (all FITC-conjugated secondary antibody. For detailed seven missense mutation altering arginine residues experimental conditions refer to (22). The images and p.A301V) with increased probability for were acquired using an Olympus IX51 (Olympus, mutational events (32). Nine missense mutations, Hamburg, Germany) inverted research microscope three nonsense mutations, one intronic mutation, × and a 100 oil immersion objective for appropri- the duplication, and the three deletions have not ate magnification. been published so far (Table 2).

Results Sequence alterations in the GLB1 studies Sixteen GM1 gangliosidosis patients of European, Overexpressed wildtype ß-Gal in COS-1 cells ± Central American, African, and Middle Eastern resulted in enzymatic activity of 556.6 123.4 origin were studied (Table 1). We found 12 nmol/h/mg protein, whereas the basal activity of compound heterozygous and four homozygous COS-1 cells, transfected with pcDX vector devoid patients; patients 1 and 5 are of consanguineous of insert, was 71.9 ± 14.9 nmol/h/mg protein. parents. Out of 28 detected sequence alterations Four mutations, c.335A>C (p.H112P), c.380G>A in the genomic DNA of these patients 18 were (p.C127Y), c.542T>A (p.I181K), and c.1343A>T missense mutations, 3 nonsense mutations, 2 (p.D448V), did not display any ß-Gal activity intronic mutations, 1 duplication, 3 deletions, upon expression in COS-1 cells (Table 3). The and 1 insertion (Table 2). They are located in other mutations expressed, c.146G>A (p.R49H), several exons among the entire GLB1 gene, c.217A>G (p.K73E), c.319T>C (p.F107L), with a cluster in exon 2 of 15.6%. As in the c.688T>C (p.C230R), c.902C>T (p.A301V), and duplication, the deletions and the insertion, only c.1166T>C (p.L389P), showed enzyme activities one nucleotide was found to be changed; they equal to or below 6.1% residual ß-Gal wildtype all result in a frameshift and consequently in a activity (Table 3). Besides p.C230R, all the other premature stop codon 5 to 23 codons downstream mutations with detectable enzyme activity were

Table 3. ß-Galactosidase activity in transiently transfected COS-1 cells

ß-Galactosidase activity Percentage ß-Hexosaminidase (nmol/h/mg of wildtype activity (nmol/h/mg protein) activity protein) Number of replicas

Wildtype 556.6 ± 123.4 100.0 637.8 ± 90.7 17 Basal 71.9 ± 14.9 0 605.1 ± 64.2 17 p.R49H 101.4 ± 19.9 6.1 623.5 ± 102.0 6 p.K73E 88.2 ± 21.1 3.4 598.4 ± 71.3 7 p.F107L 77.8 ± 14.9 1.2 549.4 ± 126.5 5 p.H112P 67.9 ± 14.4 0 594.2 ± 74.0 6 p.C127Y 63.4 ± 8.7 0 546.7 ± 125.5 6 p.I181K 67.4 ± 15.4 0 647.3 ± 63.7 5 p.C230R 89.7 ± 17.7 3.7 606.0 ± 120.5 8 p.A301V 79.2 ± 19.5 1.5 680.9 ± 102.3 6 p.L389P 74.2 ± 17.8 0.5 624.4 ± 162.2 6 p.D448V 59.4 ± 10.5 0 572.9 ± 110.3 6 p.W582X 62.8 ± 21.6 0 584.45 ± 135.3 5 p.E620X 68.0 ± 27.1 0 487.91 ± 124.6 5 p.K659X 567.2 ± 89.1 102.2 549.98 ± 155.1 5

5 Hofer et al.

Fig. 1. Western blot analysis of mutations expressed in COS-1 cells. In each lane, 20 μg of total protein was loaded. Control was transfection of the pcDX vector without insert. The 85 kDa precursor band is detected in all expression products, including all mutated as well as the wildtype allele. identified in juvenile and adult patients. The signif- Immunostaining of fibroblasts icant reduction of enzyme activity in the assayed In accordance to the presence of ß-Gal precursor mutations is in consistence with the pathogenic in Western blots from overexpressing COS-1 cells, character of the sequence alterations. Among the ß-Gal protein was detected in all six GM1 gan- expressed artificial nonsense mutations in exon 16, gliosidosis fibroblast lines as well as in the control p.W582X and p.E620X did not show any residual cell line (Fig. 3). The enzyme was distributed in enzyme activity, whereas p.K659X displayed the punctuate organelles over the entire cell body and wildtype activity (Table 3). there was no difference in fluorescence distribu- tion between healthy controls and the GM1 cell Western blot analysis lines. As recently shown with proven transport- deficient mutants, one has to conclude that the From the detection of ß-Gal precursor proteins mutated proteins were transported into lysosome- upon Western blotting of the overexpressing COS- related membrane systems (22). 1 cell homogenates, we conclude that stable ß-Gal precursors were formed (Fig. 1). Western blotting of fibroblast cell homogenates of affected patients Discussion show a faint 64 kDa mature ß-Gal band in the patients compared to the control cells (Fig. 2). In this paper, we describe 18 missense muta- The 85 kDa precursor band is present in all tions, 3 nonsense mutations, 2 intronic mutations, patient homogenates comparable to the control 3 small deletions, a small insertion, and a dupli- cell line. Several faint degradation bands are cation detected in the genomic DNA of 16 GM1 visible in all cell lines tested. Infantile patient gangliosidosis patients (Table 2). Nine of the mis- 3 (p.I181K/p.I181K) shows a considerable strong sense mutations, the three nonsense mutations, the degradation band at approximately 40 kDa. The deletions, one of the intronic mutations, and the PPCA antibody recognized the precursor band of duplication have not been published so far. Over 54 kDa and the catalytically active 32 kDa and 100 mutations of the GLB1 gene have to date been 20 kDa chains. Although the precursor protein is shown to cause a broad spectrum of clinical man- merely present in the wildtype control, the smaller ifestations, ranging from cases with severe early proteins are detected in all patients comparable infantile neuronal dysfunction to attenuated skele- with the control. tal dysplasia without any affection of the central nervous system.

Fig. 2. Western blot analysis of fibroblast cell homogenates Fig. 3. Western Blot analysis of fibroblast cell homogenates using the ß-Gal antibody. In each lane, 30 μg of total protein using the protective protein/cathepsin A (PPCA) antibody. was loaded. Fibroblast homogenate of a healthy individual was In each lane, 30 μg of total protein was loaded. Fibroblast used as a control. The 85 kDa precursor protein is present in homogenate of a healthy individual was used as a control. The all cell lines tested. The mature 64 kDa protein is reduced in catalytically active 32 kDa and 20 kDa chains are present in all the patients’ cells and a strong degradation band at 40 kDa is patients tested as well as in the wildtype. The precursor protein detected in patient 3. is merely present in the wildtype control.

6 Phenotype determining alleles in GM1 gangliosidosis patients GLB1 mutations may (1) inhibit the catalytic during substrate cleavage (14,15). The common reaction, impair correct protein folding and/or allele p.W273L is supposed to result in a sta- result in the formation of unstable proteins. If ble protein with altered catalytic function, pro- located at the active site, they can result in the loss viding normal degradation to GM1 ganglioside of a catalytically essential amino acid or in a steri- but impaired cleavage of keratan sulfate (16). In cal hindrance of substrate access to the catalytic one single MBD patient (p.Y444C/p.G494S) (38), site (2). If affecting the ß-Gal surface and thus the phenotype-specific allele could so far not the multienzyme complex aggregation, enzymatic be determined, as no other reports on these activity or protein stability may be impaired. (3) If alleles are in existence. Other alleles, affecting a domain common to ß-Gal and EBP is affected, Arg201 (p.R201H) and Gly438 (p.G438E) have the resulting phenotype may reflect inhibited lyso- been detected in homozygous patients with either somal degradation and impaired assembly of extra- juvenile/adult GM1 or MBD (23,28,34,40,42) and cellular matrix and/or processes related to signal therefore cannot be clearly assigned to a phenotype. transduction (33–35). (4) Due to the influence of Among others (p.I51T (19); p.L155R (26); modifier , even identical genotypes may not p.T420K (42)), p.T82M has repeatedly been shown always result in the same phenotypes among dif- to result in adult GM1, independent of the coun- ferent patients (36). terallele. So far, five heteroallelic p.T82M patients Despite this complex situation, the available were detected. They all presented with a delay results on clinical genetics of the GLB1 gene of speech development and clumsiness between provide increasing empirical evidence on corre- 2 and 4 years of age, subsequent development of lations between single alleles and distinct phe- spondyloepiphyseal dysplasia and decreased ß-Gal notypes, either due to homoallelic occurrence in activity in fibroblasts or leukocytes between 2.6% single patients or due to the presence of counteral- and 3.4% of controls (21,22). These five patients leles resulting in truncated dysfunctional proteins. appear to be surprisingly uniform in onset and out- As summarized in Table 4, the results of this paper come, thus confirming a phenotype determining correlate well with previous data and may provide effect of p.T82M. a basis for further molecular investigations. Homoallelic patients Known specific alleles GM1 patients 1, 3, 5, and 7 were homoallelic Few mutations in the GLB1 gene could so far with their mutations directly related to infantile be clearly related to a specific GM1 or MBD (c.245+1G>A, p.I181K, and p.C230R) and juve- phenotype. Three codons were found to be cru- nile GM1 (p.C127Y), respectively. Homozygous cial for MBD: (i) Tyr83 (p.Y83C/p.D441N (38); patient 5 (p.C230R) appears to be of particular p.Y83H/p.R482C (39)); (ii) Tyr500 (p.N484K/ notice. Lysosomal storage was suspected at birth p.T500A (40); p.L173P/p.T500A (38); p.Q408P/ due to storage cells in the placenta rather than p.T500A (22)), and (iii) Trp273 (p.W273L/ due to typical GM1 symptoms and final diagno- p.W273L (14); p.W273R/p.H281Y (41)). sis achieved at 2 months of age. His early and Contrary to the recent in silico-model (12), pre- severe neurological symptoms (severe spasticity, vious evidence suggests that codon p.W273 is cherry-red spots) and poor cognitive outcome until involved in the catalytical function and may stabi- to his current age of 2 were consistent with a clas- lize the galactose residue in the catalytic pocket sification as infantile GM1 (1) despite a proven

Table 4. Phenotype determining alleles in heteroallelic GM1 patients in this paper

Allele Genotype Phenotype Considerations p.R49H p.R49H/p.G526GfsX5 Adult GM1 The counter allele results in a truncated protein p.K73E p.K73E/p.R148C Adult GM1 Infantile homozygous patient p.R148C/p.R148C reported (37) p.F107L p.F107L/p.D448GfsX8 Juvenile GM1 The counter allele results in a truncated protein p.T82M p.T82M/p.H112P Adult GM1 Adult heterozygous patient p.T82M/c.75+2 75+3insT reported (21) p.L155R p.W65X/p.L155R Adult GM1 Adult homozygous patient p.L155R/p.L155R reported (26) p.L155R/p.R299VfsX5 Adult GM1 p.R201H p.I55FfsX16/p.R201H Juvenile GM1 p.R201H associated with GM1 2/3 (15,30,31) and MDB (38) p.R208C p.W161X/p.R208C Infantile GM1 Infantile homozygous patient p.R208C/p.R208C reported (27) p.R442Q p.F357L + p.K359KfsX23 /p.R442Q Adult GM1 Adult heterozygous patient p.T329A/p.R442Q reported (28) p.D448V p.W273X/p.D448V Infantile GM1 The nonsense mutation results in a truncated, nonfunctional protein

7 Hofer et al.

Fig. 4. Immunolocalization of ß-Gal in cultured fibroblasts. A, Wildtype; B, patient 3, genotype p.I181K/p.I181K; C, patient 4, genotype p.W161X/p.R208C; D, patient 5, genotype p.C230R/p.C230R; E, patient 8, genotype p.F107L/p.D448GfsX8; F, patient 10, genotype p.A301V/p.L389P; G, patient 14, genotype p.W65X/p.L155R. ß-Gal distribution in all patients’ fibroblast cells is comparable to the control cell line.

8 Phenotype determining alleles in GM1 gangliosidosis patients lack of cardiac dysfunction, the absence of hep- results from two homozygous patients with iden- atosplenomegaly and only minor symptoms of tical onset of disease during the second half-year bone disease. Some of the visceral symptoms may of life (27). have been ameliorated by stem cell transplanta- As listed in Table 4, several other mutations tion at 4 months of age. However, ß-Gal activity can be related to juvenile and adult GM1 by in fibroblasts (3.5% of wildtype) and the activity similar considerations: p.R201H is well-known to of mutant precursors after overexpression in COS- protect patients from severe GM1 (26,28,34,38) 1 cells (3.7% of wildtype) were similar to GM1 and consistently patient 9, genotype p.I55FfsX16/ type 2 and 3 and may account for the relatively p.R201H, was manifested as juvenile GM1. In mild organ affection and minor dysmorphism. analogy, p.F107L, displaying 1.2% of normal ß- As p.C230R is located in exon 6, which is Gal activity upon overexpression, has to be deter- skipped in the alternatively spliced GLB1 gene minant for juvenile patient 8, as the counter allele product, EBP, it can be supposed to be fully p.D448GfsX8 results in a truncated ß-Gal pro- functional. In accordance with previous reports, tein. p.L155R, previously found in a homozygous this would imply normal elastogenesis and con- patient classified as GM1 type 3 with gait and sequently normal cardiac function (33). Similar speech disturbances at the age of 6 (26), has to to other atypical phenotypes of the GLB1 gene, be dominant in the patients 14 (p.W65X/p.L155R) knowledge on the consequences of the C230R and 15 (p.L155R/p.R299VfsX5). allele on protein function, including the degrada- Mutation p.R442Q has so far been found in tion of natural substrates is required for an expla- patient 16 and in another adult patient (28), nation of the relatively high residual activity of implicating a significance for adult GM1. Despite mutant ß-Gal in the severely affected patient 5. single reports on neighboring missense mutations resulting in the severe phenotype, a conspicuous clustering of mutations associated with mild GM1 or MBD in exon 13, with minor impact on ß-Gal Heteroallelic patients function can be proposed (23,26,38,40). As little was known on the potential residual Mutation p.R49H reveals 6.1% of wildtype activity of truncated ß-Gal mutants, three artifi- activity in COS-1 cells and, due to the nonsense cial nonsense mutations (p.W582X, p.E620X and mutation on the counter allele, has to be considered p.K659X) were introduced in exon 16 (Table 4). as the phenotype-specific allele in adult patient In accordance with previous results on the impor- 11, genotyped p.R49H/p.G526GfsX5. Patient 12, tance of the C-terminus for ß-Gal function (31), p.K73E/p.R148C, presented at the age of 16 was p.W582X and p.E620X did not result in any resid- classified adult GM1. As a homozygous case ual activity. Merely, p.K659X, located 19 amino for p.R148C presented with infantile GM1 (37), acids prior to the natural stop codon, had nor- p.K73E has to be phenotype-determinant. mal catalytic activity. In exon 16, no naturally Juvenile patient 10 carries the mutations occurring termination mutation has so far been p.A301V and p.L389P, resulting in 1.5% and identified and p.W576X was the mutation located 0.5% of wildtype activity upon COS-cells expres- closest to the natural stop codon (26). Thus, all sion, respectively. Both mutations have not been detected nonsense mutations resulting in shorter detected so far and no conclusions on the truncated proteins than the introduced termina- phenotype-determinant allele can be drawn. tion codons have to be devoid of ß-Gal activity. Taken together, 17 novel sequence alterations in the GLB1 gene could be described. Upon In heteroallelic patients with nonsense mutations, comparison of our results with previous data, the counterallele has thus to be determinant for phenotype specificities in heteroallelic genotypes the phenotype, as illustrated in the infantile GM1 were proposed for nine GLB1 mutations. Future patients 4 and 6 bearing the missense mutations observations, biochemical experiments and data on p.R208C and p.D448V, respectively, besides non- molecular structure of the ß-Gal protein have to sense mutations on the other allele. Expression confirm their predictive value for GM1 gangliosi- products of p.D448V in COS-1 cells were com- dosis patients. pletely devoid of enzymatic activity. Furthermore, patient 6 (p.W273X/p.D448V) would have no functional ß-Gal enzyme at all, while for patient Acknowledgements 4 (p.W161X/p.R208C), any potential ß-Gal activ- This work was supported by the Austrian Ministry of Science and ity would be attributable to p.R208C, similar to Research (single project GZ 200.156/2-VI/1a/2006 entitled ‘Inves- a homozygous patient. This is consistent with tigations for the Early Diagnosis and Prognosis of the Phenotype

9 Hofer et al. in Lysosomal Storage Diseases’) and the Austrian Research Soci- 13. Paschke E, Milos I, Kreimer-Erlacher H et al. Mutation anal- ety for Mucopolysaccharidosis and Related Diseases. The authors yses in 17 patients with deficiency in acid beta-galactosidase: are indebted to A. Morrone, University of Florence, Italy, and three novel point mutations and high correlation of mutation A. d’Azzo, St. Jude Children’s Research Hospital in Memphis, TN, W273L with Morquio disease type B. Hum Genet 2001: 109: for helpful discussions and provision of the GLB1 pcDX-X cDNA 159–166. and the α85 antihuman ß-galactosidase antibody. Careful reports on 14. Oshima A, Yoshida K, Shimmoto M et al. Human beta- clinical data were provided by D. Mansour, Department of Contra- galactosidase gene mutations in morquio B disease. Am J Hum ception and Sexual Health, Graingerville Clinic, Newcastle General Genet 1991: 49: 1091–1093. Hospital, UK, K. Tylee, Willink Biochemical Genetics Unit, Royal 15. Callahan JW. Molecular basis of GM1 gangliosidosis and Manchester Children’s Hospital, Manchester, UK and F.S. Ezgu,¨ Morquio disease, type B. Structure-function studies of Gazi University Hospital Ankara, Turkey. We gratefully acknowl- lysosomal beta-galactosidase and the non-lysosomal beta- edge the technical assistance of Bettina Pabst and Tania Skachkova galactosidase-like protein. Biochim Biophys Acta 1999: 1455: and the patients’ cooperation for providing fibroblasts cultures. 85–103. 16. Okumiya T, Sakuraba H, Kase R et al. Imbalanced substrate specificity of mutant beta-galactosidase in patients with Conflicts of interest Morquio B disease. Mol Genet Metab 2003: 78: 51–58. All authors state that there are no actual or poten- 17. Paschke E, Kresse H. Morquio disease, type B: activation of GM1-beta-galactosidase by GM1-activator protein. Biochem tial conflicts of interest in the present study, includ- Biophys Res Commun 1982: 109: 568–575. ing any financial, personal or other relationships 18. Oshima A, Yoshida K, Itoh K et al. Intracellular processing with other people or organizations. and maturation of mutant gene products in hereditary beta- galactosidase deficiency (beta-galactosidosis). Hum Genet 1994: 93: 109–114. References 19. Yoshida K, Oshima A, Shimmoto M et al. Human beta- 1. Suzuki Y, Oshima A, Nanba E. ß-Galactosidase deficiency (ß- galactosidase gene mutations in GM1-gangliosidosis: a com- galactosidosis): GM1 gangliosidosis and Morquio B disease. mon mutation among Japanese adult/chronic cases. Am J Hum In: Scriver CR, Beaudet AL, Sly WS et al., eds. The metabolic Genet 1991: 49: 435–442. and molecular basis of inherited disease. New York: McGraw- 20. Kaye EM, Shalish C, Livermore J et al. beta-Galactosidase Hill Inc, 2001: 3775–3809. gene mutations in patients with slowly progressive GM1 2. Yamamoto Y, Hake CA, Martin BM et al. Isolation, charac- gangliosidosis. J Child Neurol 1997: 12: 242–247. terization, and mapping of a human acid beta-galactosidase 21. Chakraborty S, Rafi MA, Wenger DA. Mutations in the lyso- cDNA. DNA Cell Biol 1990: 9: 119–127. somal beta-galactosidase gene that cause the adult form of 3. Morreau H, Galjart NJ, Gillemans N et al. Alternative splicing GM1 gangliosidosis. Am J Hum Genet 1994: 54: 1004–1013. of beta-galactosidase mRNA generates the classic lysosomal 22. Hofer D, Paul K, Fantur K et al. GM1 gangliosidosis and enzyme and a beta-galactosidase-related protein. J Biol Chem Morquio B disease: expression analysis of missense mutations 1989: 264: 20655–20663. affecting the catalytic site of acid beta-galactosidase. Hum 4. Hinek A, Rabinovitch M, Keeley F et al. The 67-kD elastin/ Mutat 2009: 30: 1214–1221. laminin-binding protein is related to an enzymatically inactive, 23. Roze E, Paschke E, Lopez N et al. Dystonia and parkinsonism alternatively spliced form of beta-galactosidase. J Clin Invest in GM1 type 3 gangliosidosis. Mov Disord 2005: 20: 1993: 91: 1198–1205. 1366–1369. 5. Privitera S, Prody CA, Callahan JW et al. The 67-kDa enzy- 24. Georgiou T, Drousiotou A, Campos Y et al. Four novel muta- matically inactive alternatively spliced variant of beta- tions in patients from the Middle East with the infantile form galactosidase is identical to the elastin/laminin-binding protein. of GM1-gangliosidosis. Hum Mutat 2004: 24: 352. J BiolChem 1998: 273: 6319–6326. 25. Hilson WL, Okamura-Oho Y, Zhang S et al. Novel mis- 6. Hinek A, Pshezhetsky AV, Callahan JW. A cell surface com- sense mutations in ß-galactosidase that results in GM1- plex of spliced ß-galactosidase variant protein, neuraminidase gangliosidosis. Am J Hum Genet 1994: 55 (S3): 1302. and protectice protein controls elastic fiber assembly. J Inherit 26. Santamaria R, Blanco M, Chabas A et al. Identification of 14 Metab Dis 1998: 21: 112. novel GLB1 mutations, including five deletions, in 19 patients 7. Ostrowska H, Krukowska K, Kalinowska J et al. Lysosomal with GM1 gangliosidosis from South America. Clin Genet high molecular weight multienzyme complex. Cell Mol Biol 2007: 71: 273–279. Lett 2003: 8: 19–24. 27. Boustany RM, Qian WH, Suzuki K. Mutations in acid beta- 8. Pshezhetsky AV, Ashmarina M. Lysosomal multienzyme com- galactosidase cause GM1-gangliosidosis in American patients. plex: biochemistry, genetics, and molecular pathophysiology. Am J Hum Genet 1993: 53: 881–888. Prog Nucleic Acid Res Mol Biol 2001: 69: 81–114. 28. Caciotti A, Donati MA, Bardelli T et al. Primary and sec- 9. Bagshaw RD, Zhang S, Hinek A et al. Novel mutations (Asn ondary elastin-binding protein defect leads to impaired elasto- 484 Lys, Thr 500 Ala, Gly 438 Glu) in Morquio B disease. genesis in fibroblasts from GM1-gangliosidosis patients. Am Biochim Biophys Acta 2002: 1588: 247–253. J Pathol 2005: 167: 1689–1698. 10. Caciotti A, Donati MA, Boneh A et al. Role of beta- 29. Silva CM, Severini MH, Sopelsa A et al. Six novel beta- galactosidase and elastin binding protein in lysosomal and galactosidase gene mutations in Brazilian patients with GM1- nonlysosomal complexes of patients with GM1-gangliosidosis. gangliosidosis. Hum Mutat 1999: 13: 401–409. Hum Mutat 2005: 25: 285–292. 30. Lowry OH, Rosebrough NJ, Farr AL et al. Protein measure- 11. Hoogeveen AT, Graham-Kawashima H, d’Azzo A et al. Pro- ment with the Folin phenol reagent. J Biol Chem 1951: 193: cessing of human beta-galactosidase in GM1-gangliosidosis 265–275. and Morquio B syndrome. J Biol Chem 1984: 259: 1974–1977. 31. van der Spoel A, Bonten E, d’Azzo A. Processing of lysoso- 12. Morita M, Saito S, Ikeda K et al. Structural bases of GM1 mal beta-galactosidase. The C-terminal precursor fragment is gangliosidosis and Morquio B disease. J Hum Genet 2009: an essential domain of the mature enzyme. J Biol Chem 2000: 54: 510–515. 275: 10035–10040.

10 Phenotype determining alleles in GM1 gangliosidosis patients

32. Holliday R, Pugh JE. DNA modification mechanisms and gene 38. Santamaria R, Chabas A, Coll MJ et al. Twenty-one novel activity during development. Science 1975: 187: 226–232. mutations in the GLB1 gene identified in a large group 33. Hinek A, Zhang S, Smith AC et al. Impaired elastic-fiber of GM1-gangliosidosis and Morquio B patients: possible assembly by fibroblasts from patients with either Morquio B common origin for the prevalent p.R59H mutation among disease or infantile GM1-gangliosidosis is linked to deficiency gypsies. Hum Mutat 2006: 27: 1060–1071. in the 67-kD spliced variant of beta-galactosidase. Am J Hum 39. Ishii N, Oohira T, Oshima A et al. Clinical and molecular Genet 2000: 67: 23–36. analysis of a Japanese boy with Morquio B disease. Clin Genet 34. Morrone A, Bardelli T, Donati MA et al. beta-galactosidase 1995: 48: 103–108. gene mutations affecting the lysosomal enzyme and the elastin- 40. Hinek A, Zhang S, Smith AC et al. Impaired elastic-fiber binding protein in GM1-gangliosidosis patients with cardiac assembly by fibroblasts from patients with either Morquio B involvement. Hum Mutat 2000: 15: 354–366. disease or infantile GM1-gangliosidosis is linked to deficiency 35. Caciotti A, Donati MA, Bardelli T et al. Primary and sec- in the 67-kD spliced variant of beta-galactosidase. Am J Hum ondary elastin-binding protein defect leads to impaired elasto- Genet 2000: 67: 23–36. genesis in fibroblasts from GM1-gangliosidosis patients. Am 41. Gucev ZS, Tasic V, Jancevska A et al. Novel beta- J Pathol 2005: 167: 1689–1698. galactosidase gene mutation p.W273R in a woman with 36. Koprivica V, Stone DL, Park JK et al. Analysis and classifi- mucopolysaccharidosis type IVB (Morquio B) and lack of cation of 304 mutant alleles in patients with type 1 and type response to in vitro treatment of her skin fibroblasts. 3 Gaucher disease. Am J Hum Genet 2000: 66: 1777–1786. Am J Med Genet A 2008: 146A: 1736–1740. 37. Caciotti A, Donati MA, Procopio E et al. GM1 gangliosidosis: 42. Santamaria R, Chabas A, Callahan JW et al. Expression and molecular analysis of nine patients and development of an RT- characterization of 14 GLB1 mutant alleles found in GM1- PCR assay for GLB1 gene expression profiling. Hum Mutat gangliosidosis and Morquio B patients. J Lipid Res 2007: 48: 2007: 28: 204–215. 2275–2282.

11