HUMAN MUTATION Mutation in Brief #986 (2007) Online

MUTATION IN BRIEF

© 2007 WILEY-LISS, INC.

Received 24 March 2007; accepted revised manuscript 11 August 2007.

Molecular Analysis of SUMF1 Mutations: Stability and Residual Activity of Mutant Formylglycine-Generating Enzyme Determine Disease Severity in Multiple Sulfatase Deficiency Lars Schlotawa1, Robert Steinfeld1, Kurt von Figura2, Thomas Dierks3*, and Jutta Gärtner1 1Department of Pediatrics and Pediatric Neurology, Georg August University Göttingen, Göttingen, Germany; 2Department of Biochemistry II, Georg August University Göttingen, Göttingen, Germany; 3Department of Chemistry, Biochemistry I, Bielefeld University, Bielefeld, Germany

*Correspondence to: Thomas Dierks, Department of Chemistry, Biochemistry I, Bielefeld University, Universitätsstr. 25, 33615 Bielefeld, Germany; Tel.: +49 521 106 2092; Fax: +49 521 106 6014; E-mail: thomas.dierks@uni-bielefeld.de Grant sponsors: Deutsche Forschungsgemeinschaft; Fonds der Chemischen Industrie; Shire Human Genetic Therapies, Inc. Communicated by Johannes Zschocke

Multiple Sulfatase Deficiency (MSD) is a rare inborn autosomal-recessive disorder, which mainly combines clinical features of metachromatic leukodystrophy, mucopolysaccharidosis and X-linked ichthyosis. The clinical course ranges from neonatal severe to mild juvenile cases. MSD is caused by mutations in the SUMF1 gene encoding the formylglycine-genera-ting enzyme (FGE). FGE posttranslationally activates sulfatases by generating formylglycine in their catalytic sites. We analyzed the functional consequences of missense mutations p.A177P, p.W179S, p.A279V and p.R349W with regard to FGE’s subcellular localization, enzymatic activity, protein stability, intracellular retention and resulting sulfatase activities. All four mutations did not affect localization of FGE in the endoplasmic reticulum of MSD fibroblasts. However, they decreased its specific enzymatic activity to less than 1% (p.A177P and p.R349W), 3% (p.W179S) or 23% (p.A279V). Protein stability was severely decreased for p.A279V and p.R349W, and almost comparable to wild type for p.A177P and p.W179S. The patient with the mildest clinical phenotype carries the mutation p.A279V leading to decreased FGE protein stability, but high residual enzymatic activity and only slightly redu-ced sulfatase activities. In contrast, the most severely affected patient carries the mutation p.R349W leading to drastically decreased protein stability, very low residual enzymatic activity and considerably reduced sulfatase activities. Our functional studies provide novel insight into the molecular defect underlying MSD and reveal that both residual enzyme activity and protein stability of FGE contribute to the clinical phenotype. The application of improved functional assays to determine these two molecular parameters of FGE mutants may enable the prediction of the clinical outcome in the future. © 2007 Wiley-Liss, Inc.

KEY WORDS: multiple sulfatase deficiency; formylglycine generating enzyme; SUMF1; genotype-phenotype; sulfatases

INTRODUCTION

Multiple sulfatase defiency (MSD, MIM# 272200) is a rare autosomal-recessive inborn error characterized by a dysfunction of all sulfatases. Due to the catabolic function of at least six sulfatases in lysosomes, MSD patients

DOI: 10.1002/humu.9515

2 Schlotawa et al.

show massive lysosomal storage of sulfated glycosaminoglycans and sulfolipids. The clinical phenotype is highly variable ranging from a rapidly progressive neonatal form to milder juvenile ones. The clinical symptoms overlap with those described for known single sulfatase deficiencies like metachromatic leukodystrophy, mucopoly-saccharidoses, X-linked ichthyosis and chondrodysplasia punctata type I [Hopwood and Ballabio, 2001]. The neonatal form is characterized by a coarse face, cataract and hydrocephalus resembling mucopolysaccharidosis, whereas later onset forms are closer to infantile or juvenile metachromatic leukodystrophy with often normal early development followed by psychomotor impairment and a neurodegenerative course [Burch et al., 1986; Vamos et al., 1981; Perlmutter-Cremer et al., 1981; Kepes et al., 1988].

The responsible gene was discovered recently and termed sulfatase modifying factor 1 (SUMF1, MIM# 607939) [Dierks et al., 2003; Cosma et al., 2003]. It encodes the formylglycine (FGly) generating enzyme (FGE), a key enzyme required for posttranslational activation of sulfatases [Dierks et al., 2003]. So far, 17 human genes encoding sulfatases have been identified and eight of them are associated with human diseases presenting with clearly defined clinical pictures [Diez-Roux and Ballabio, 2005]. All sulfatases share amino acid homology in their active sites including a cysteine residue in the core-motif [Dierks et al., 1999; Sardiello et al., 2005]. This so-called sulfatase signature determines the cysteine to be posttranslationally modified to FGly by FGE in the endoplasmic reticulum (ER) [Dierks et al., 1997; Dierks et al., 1999], which is a strict precondition to become a catalytically active sulfatase [Schmidt et al., 1995].

So far, 30 mutations in SUMF1 have been described in MSD patients including nine nonsense and 21 missense mutations [Dierks et al., 2003; Cosma et al., 2003; Cosma et al, 2004; Dierks et al., 2005; Diaz-Font et al., 2005]. Mature FGE is made up of a 341-amino acid polypeptide, which adopts a single domain monomeric structure with a unique fold [Dierks et al., 2005]. As predicted from crystal structures, SUMF1 missense mutations found in MSD patients may lead to FGE instability, interfere with substrate binding or catalytic activity [Dierks et al., 2005].

In a first attempt to study the functional consequences of SUMF1 mutations in vitro, Cosma and co-workers transiently co-expressed mutant FGE with sulfatases. They found that the majority of SUMF1 missense mutations lead to impaired activity of all nine sulfatases tested [Cosma et al., 2004]. Two missense mutations led to a high residual activity of individual sulfatases, suggesting that SUMF1 missense mutations may not affect each sulfatase to the same extent. When comparing the clinical phenotype of 20 MSD patients with known missense mutations to the level of residual sulfatase activities measured in the co-expression experiments, they did not find a correlation [Cosma et al., 2004].

In this study we analyzed the direct functional consequences of MSD causing SUMF1 mutations on FGE by determining its subcellular localization, protein stability and residual enzyme activity. The molecular and biochemical data are related to the clinical phenotype of patients with MSD.

MATERIALS AND METHODS

Patients and patient cells Clinical data were obtained from published case reports, personal communications and patient data sheets and

are summarized in Table 1. Five of the patients carry homozygous SUMF1 missense mutations, namely c.1045C>T (p.R349W) in three patients (#1-3), c.836C>T (p.A279V) (#6) and c.536 G>C (p.W179S) (#5). Patient 4 has the SUMF1 mutation c.529G>C (p.A177P) in heterozygous form in combination with a severe frameshift/ truncation mutation c.748delC (p.L250fs); GenBank reference sequence AY208752.1 with the A of the ATG translation start codon referred to as nucleotide +1. All patients were born after uneventful pregnancies at term, one showed an intrauterine growth retardation. Psychomotor retardation and/or first signs of a neurodegenerative disorder became evident within the first two years of life. All patients developed ichthyosis and skeletal changes, but with a highly variable age of onset. Organomegaly was obvious in four patients, not defined in one patient and absent in another. Dysmorphic features were missing in one patient and present with variable expression in the others. None of the patients did reveal corneal clouding or hydrocephalus.

Skin fibroblasts were available for all patients, namely G1608/97 (patient 1), G878/93 (#2), G3051/03 (#3), G3052/03 (#4), G3069/03 (#5) and G3087/03 (#6). We also included one fetal cell line (G3088/03) in our study.

FGE expression FGE cDNA (AY208752.1) served as template for an add-on PCR reaction using primers FGE_SP_EcoRI

(GGAATTCGGGACAACATGGCTGCG) and FGE_XbaI_nc (GCTCTAGACCTTGGTTGTCAGTCCATGG).

Molecular Analysis of SUMF1 Mutations in Multiple Sulfatase Deficiency 3

The PCR product was cloned into the pSB4.7pA expression vector (Shire Human Genetic Therapies, Inc., Cambridge, MA) downstream of its CMV promotor. The pSB-FGE plasmid was used as template for site-directed mutagenesis according to the QuikChange Protocol (Stratagene) with Pfu polymerase and complementary primer pairs (coding sequence only): FGE_A177Pc 5’-GGCAGTTGCAGCTCCTCCCTGGTGG-3’, FGE_W179Sc 5’-CAGCTGCTCCCTCGTGGTTACCTGTG-3’, FGE_A279Vc 5’-CTTCCAAGGAACTGTGCCTGTTGATGCC-3’, FGE_R349Wc 5’-CGCTGTGCTGCTTGGAGCCAGAACAC-3’. DpnI-treated products were amplified in E. coli. All constructs were fully sequenced to preclude any PCR error.

HT-1080 cells were stably transfected with these plasmids as described earlier [Preusser-Kunze et al., 2005].

Western blot analysis Equivalent aliquots of medium (40 µl) or of cell lysate, containing 50 or 100 µg cell protein, were analyzed

using a polyclonal anti FGE antibody [Dierks et al., 2005; Preusser-Kunze et al., 2005]. ECL signals were detected on a LAS3000 system (Fuji) and quantified using the AIDA software package (Raytest). Endogenous FGE in MSD fibroblasts was detected using ECL femto substrate and enhancer diluted 1:10 in ECL solution (Pierce).

Immunofluorescence microscopy

To detect endogenous FGE or mutant FGE in stably transfected HT-1080 cells or human skin fibroblasts, cells were grown on coverslips for 24 hours and analyzed by indirect immunofluorescence as previously described [Preusser-Kunze et al., 2005]. Co-localization studies were done using FGE antiserum (see above) and monoclonal antibodies against organelle markers, namely PDI (from Stressgene), LAMP1 (from Abcam) and Golgi protein 58K (Abcam). Confocal images were taken on a Leica TCS SP2 AOBS laser scan microscope.

Purification of FGE protein Stably overexpressing HT-1080 cells were grown in medium without fetal calf serum, which was collected

every 48 hours and subjected to ammonium sulfate precipitation [Preusser-Kunze et al., 2005]. The precipitate was reconstituted in and dialyzed against buffer A (10 mM Tris-HCl, pH 8.0, 100 mM NaCl, 2.5 mM dithiothreitol). The dialyzed material was applied to a MonoQ PC 1.6/5 anion exchange column (GE Healthcare). After washing thoroughly, proteins were eluted by a 1 ml linear gradient of buffer B (10 mM Tris- HCl, pH 8.0, 100 mM NaCl, 2.5 mM dithiothreitol, 1 M KCl) ranging from 0 to 37.5% (3.75% per minute). Elution fractions containing FGE, as tested by Western blotting, were pooled, lyophilized, reconstituted in 100 µl buffer A and used for FGE quantification (on calibrated Western blots) and enzyme assays. For purification of intracellular FGE proteins, cells were grown to near confluency, trypsinized and washed with phosphate-buffered saline. After reconstitution in buffer A (supplemented with a protease inhibitor cocktail), sonification and centrifugation, the supernatant was loaded on the MonoQ anion exchange column following the procedure mentioned above.

Enzyme assays Sulfatase activities were measured from whole cell lysates as described before [Rommerskirch et al., 1992;

Voznyi et al., 2001]. FGE activity assays were performed as described earlier using the synthetic ASA-derived peptide P23 as substrate at 200 nM concentration [Dierks et al., 2003; Preusser-Kunze et al., 2005]. After incubation for 60 and 120 minutes at 37°C, FGly formation was determined using MALDI-TOF mass spectrometry. The specific activity was calculated by relating to FGE protein amount determined by Western blotting. FGE amounts were adjusted in order not to exceed 50% turnover of P23 in 30 min at 37°C.

Metabolic labelling and immunoprecipitation Cells were grown in 3 cm plates to 100% confluency, starved for 90 minutes in 2 ml medium depleted of

methionine and cysteine and pulsed with 100 µCi/ ml 35S-labeled methionine/ cysteine for another 90 minutes. Medium was changed to 2 ml medium with 10% fetal calf serum and supplemented with 200 µg/ml methionine and 100 µg/ml cysteine for 15 minutes. Cells and medium were harvested either directly or after another exchange of medium and growth for 6 hours (chase). Immunoprecipitation of FGE was performed with an FGE polyclonal antiserum according to a protocol described earlier [Gieselmann et al., 1992]. Proteins were solubilized in SDS-PAGE loading buffer and subjected to gel electrophoresis. Dried gels were analyzed by phosphorimaging (BAS 1000, Fuji) and densitometric quantification using the MacBAS software (Raytest).

4 Schlotawa et al.

RESULTS

SUMF1 mutations, associated clinical phenotypes and patient classification Four different disease causing human SUMF1 missense mutations, which lead to various clinical MSD

phenotypes, are included in this study (Table 1). Three of the four mutant alleles had been identified as homozygous mutations in MSD patients. In one patient only the mutant allele p.A177P, but not the frameshift/ truncation allele p.L250fs, results in FGE protein expression and thus functionally represents a single mutation defect. Mutation p.A279V is frequent in patients of French origin and mutation p.R349W is frequent in patients of Turkish origin, whereas the other two mutations are private mutations.

Table 1. Description and Classification of MSD Patients

Clinical phenotype in multiple sulfatase deficiency (upper 3 lines) and in MSD patients studied (#1-6). Patient 3 is included in Cosma et al. [2003]. Part of the clinical data of patient 4 and 5 are already included in Mancini et al. [2001] and Loffeld et al. [2002], respectively. The clinical data of patients 1, 2 and 6 were kindly provided by A. Kohlschütter, Hamburg (Germany), by K. Harzer, Tübingen (Germany), and by M.T. Zabot, Lyon (France), respectively. aAge at onset: +++, 0-2 years; ++, 2-4 years; +, > 4 years; nd, not defined. bNVS, neonatal very severe; LIS, late infantile severe; LIM, late infantile mild. cp.L250fs, frame shift with Leu-250 as the first affected amino acid. As the reference for cDNA numbering, the A of the ATG translation start codon (accession number AY208752.1) is referred to as nucleotide +1.

Table 1 summarizes in detail the clinical signs of the six patients (see also Materials and Methods). According

to the classification by Eto and co-workers [1987], the six patients represent late infantile disease types. To further distinguish between various clinical phenotypes we also included the age of onset of the clinical symptoms. According to this, none of the patients has the “neonatal very severe” form, three have a “late infantile severe” and three a “late infantile mild” form. The three severely affected patients all carry the homozygous SUMF1 missense mutation p.R349W. They do have a similar clinical presentation and can be clearly discriminated from the even

Molecular Analysis of SUMF1 Mutations in Multiple Sulfatase Deficiency 5

more severe neonatal form. The three milder affected patients, with patient 6 showing the mildest phenotype, do have different SUMF1 mutations (Table 1).

Expression and secretion of FGE mutants Stable transfection of human fibrosarcoma HT-1080 cells led to expression of all four FGE mutant proteins

(Fig.1A). The intracellular level of FGE-A177P and FGE-W179S was in the same range as observed for wild type FGE. FGE-A279V and FGE-R349W showed a markedly reduced intracellular level. FGE mutant proteins could also be recovered in the cell culture medium. Secretion of FGE-A177P was reduced, while FGE-W179S and -A279V showed increased secretion when compared to wild type FGE. FGE-R349W was secreted at normal levels. Wild type as well as all four mutant FGE proteins were partially processed from the full-length to an N-terminally truncated form; both forms are catalytically active, as has been shown at least for wild type FGE [Preusser-Kunze et al., 2005]. Secreted FGE was further used for activity determinations ensuring that only properly folded protein, which has passed the quality control mechanisms, is analyzed.

Residual enzymatic activity of FGE mutants

The enzymatic activity of secreted wild type and mutant FGE was measured under standard conditions using a peptide substrate derived from arylsulfatase A at a concentration of 200 nM, i.e. about 15-times above the KM for wild type FGE [Dierks et al., 2003]. Specific activities were calculated by relating measured activities to FGE protein, as quantified by Western blotting using purified FGE standards. The residual activities range from 0.5±0.1% to 22.9±4.2% of wild type FGE activity (Fig.1B). FGE-A279V showed the highest residual activity, whereas FGE-A177P and -R349W had less than 1% of wild type FGE activity. Here the catalytically inactive FGE-C336S mutant was used as a negative control (not shown, cf. [Dierks et al., 2005]). These data demonstrate that MSD causing SUMF1 missense mutations lead to the production of FGE proteins showing residual, but widely different activities in vitro.

Figure 1. Expression, secretion and residual enzymatic activity of mutant FGE. A: HT-1080 cells stably overexpressing wild type FGE or the indicated mutants were grown to confluency and maintained in fresh medium for another 24 hours. Equivalent aliquots of cell lysate (C) and medium (M) were analyzed by Western blotting. The visible bands represent intracellular full-length FGE at 41 kDa and secreted FGE (38 kDa) showing N-terminal truncations and heterogeneity in N-glycosylation, which both do not affect enzymatic activity in vitro [Preusser-Kunze et al., 2005]. The intracellular truncated form (asterisk) is supposed to result from N-terminal trimming. The data shown are representative for more than three identical experiments. B: Secreted FGE protein (wild type and mutant) was purified on a MonoQ column. Protein amounts were quantified by calibrating the Western blot signals with defined amounts of pure human FGE. Enzymatic activity was measured as FGly modification of the synthetic P23 ASA using MALDI-TOF mass spectrometry (see Methods). Specific activities were calculated and expressed as percentage of wild type FGE activity. Mean values and SD were calculated from three independent analyses.

6 Schlotawa et al.

Subcellular localization of FGE mutants Using indirect immunofluorescence, we determined the intracellular localization of FGE mutants expressed in

HT-1080 cells. Permeabilized cells were double-stained with anti FGE and anti protein disulfide isomerase or Golgi 58K antibodies as specific markers for the ER and Golgi, respectively. Untransfected cells showed a weak background signal for endogenous FGE located in the ER (Fig.2A). All four FGE mutant proteins co-localized with the ER marker. In addition, the p.A279V, p.R349W and p.W179S mutants co-localized also with the Golgi marker (Fig.2B). FGE-A279V and –R349W showed a stronger co-localization signal for Golgi than for ER. FGE-W179S predominantly localizes to the ER. These results indicate that all studied FGE mutants at least partially localize to the ER and thus fulfill a basic precondition to allow FGly generation in newly synthesized sulfatases.

Figure 2. Subcellular localization of FGE mutants expressed in HT-1080 cells. For staining of permeabilized cells, a polyclonal anti FGE antibody and specific monoclonal antibodies recognizing subcellular markers (see Methods) were used as primary antibodies. A: The merge (yellow) reveals that all FGE mutants were found in the ER. B: FGE mutants p.R349W, p.A279V and, in some cells, p.W179S could also be detected in the Golgi.

Molecular Analysis of SUMF1 Mutations in Multiple Sulfatase Deficiency 7

Stability and intracellular retention of FGE mutant proteins Apart from affecting enzymatic activity, mutations often affect the stability of the expressed proteins. In order

to study this aspect, we metabolically labeled HT-1080 cells expressing wild type and mutant FGE with [35S]methionine and [35S]cysteine. The FGE proteins were immunoprecipitated after different chase times. The total amount of cellular plus secreted FGE after six hours of chase was compared to cellular FGE present before the chase start (Fig.3). Wild type FGE is fully stable for a chase time of six hours, as about 100% of the initially labeled protein could be recovered from cells plus medium. After six hours of chase, FGE protein signals could also be detected for the four mutants. Nevertheless, when compared to wild type the signals were markedly reduced for mutant p.R349W, reduced for mutant p.A279V and slightly reduced for mutant p.W179S. In contrast, mutant p.A177P was as stable as wild type FGE with more than 90% of protein recovered at the end of the chase.

Interestingly, not only protein stability but also intracellular retention of FGE was differentially affected by the individual mutations. By comparing protein signal intensities in cells and medium it turned out that FGE-A177P shows an increased cellular retention as compared to wild type FGE, whereas mutants p.W179S and p.A279V show reduced retention. For instable FGE mutants (p.A279V and p.R349W) retention efficiency can only indirectly be deduced from secreted protein amounts, as proteasomal degradation selectively affects intracellular FGE. Therefore, retention of FGE-R349W (53% secretion) obviously is similar to retention of wild type FGE (54% secretion), though nearly no intracellular FGE-R349W protein is detectable as a consequence of degradation.

Taken together stability and retention data agree with the intra- and extracellular FGE levels shown in Fig.1, indicating that the intracellular half-life of mutants p.A279V and p.R349W is significantly shortened. In contrast, intracellular half-life is increased for mutant p.A177P. It should be noted, however, that FGE secretion is mainly due to overexpression and most likely results from saturation of the ER retention mechanism. The observed changes in cellular retention may therefore be less relevant in vivo.

Figure 3. Stability and intracellular retention of FGE mutants. Overexpressing HT-1080 cells were metabolically labeled with [35S]methionine/ [35S]cysteine for 90 minutes and then chased for six hours in label-free medium. FGE was immunoprecipitated from cell lysate or medium and analyzed by SDS-PAGE, phosphorimaging and densitometric quantification. All FGE mutants were at least partially stable and showed various degrees of intracellular retention. For estimation of retention in the case of instable mutants p.A279V and p.R349W see text. Bands and quantification results shown are representative for three independent experiments.

Analysis of endogenous FGE mutants in patient fibroblasts In a next step we wanted to verify the results obtained in HT-1080 cells for recombinant FGE mutant proteins

by studying endogenous FGE. We investigated cultured human skin fibroblasts of all six MSD patients, from whom the mutant SUMF1 alleles had been isolated (Table 1). In parallel to the p.A279V fibroblasts of patient 6, we further included p.A279V fibroblasts obtained from an MSD fetus.

8 Schlotawa et al.

Western blot analysis allowed to detect FGE mutant protein in patient cell lysates (Fig.4A). In cells carrying the p.A177P and p.W179S mutations FGE expression levels were approximately in the range of wild type levels. In contrast, cells carrying SUMF1 mutations p.A279V and p.R349W contained severely reduced FGE protein levels. In fact, Western analysis failed to detect any FGE in the p.R349W fibroblasts, though traces were detectable by immunofluorescence (see below, Fig.4B). These data are in accordance with the results from the pulse chase experiments with recombinant FGE mutants, thus confirming the stability data observed in transfected HT-1080 cells (Fig.3). By indirect immunofluorescence analysis we determined the subcellular localization of the FGE mutant proteins in the MSD patient fibroblasts (Fig.4B). For the mutants p.A177P and p.W179S the FGE proteins could be detected at about wild type levels and localizing to the ER. In contrast, for mutants p.A279V and p.R349W the fluorescence signal was extremely low, but still showing ER localization.

Figure 4. Expression and subcellular localization of endogenous FGE in MSD patient fibroblasts. A: Fibroblasts of seven MSD patients were grown to confluency. 100 µg of total cell lysate protein were analyzed by Western blotting for FGE levels. Expression of FGE mutants was in the range of wild type levels only for p.W179S and p.A177P, drastically reduced for p.A279V and not detectable for p.R349W. B: FGE was stained in permeabilized MSD fibroblasts with polyclonal antibodies (red); a monoclonal anti-PDI antibody (green) was used for ER staining. The merge (yellow) reveals co-localization of FGE with the ER marker. Very low FGE amounts were detected also for the p.R349W mutant.

Molecular Analysis of SUMF1 Mutations in Multiple Sulfatase Deficiency 9

Residual sulfatase activities in patient fibroblasts To determine the impact of the FGE mutations on the activation of cellular sulfatases, we analyzed the activities

of arylsulfatase A (ASA), arylsulfatase C (ASC) and galactose 6-sulfatase (Gal6S) in whole cell lysates from cultured MSD fibroblasts, and compared them to activities in cell lysates from control fibroblasts. In cells homozygous for mutation p.R349W the activities of all three sulfatases measured were drastically reduced, i.e. 3-12% compared to activities in wild type cells (Table 2). For cells carrying the other mutations the investigated sulfatases showed only slightly reduced activities, i.e. roughly in the range of wild type fibroblasts (Table 2). Relative ASA activities varied between 85% in cells homozygous for the p.W179S allele and 13% in heterozygous p.A177P/p.L250fs cells. ASC activity was normal in p.W179S fibroblasts and about 2-fold reduced in homozygous p.A279V cells. Gal6S activity ranged from 38-67%.

Table 2. Residual Activities of Selected Sulfatases in MSD Fibroblasts

DISCUSSION

We investigated the molecular pathology underlying FGE dysfunction in late infantile MSD. By mapping known MSD causing mutations onto the FGE crystal structure, their functional consequences had been predicted earlier [Dierks et al., 2005] (see Table 3). To verify these predictions and to find out disease causing differences to wild-type FGE with respect to expression/stability, localization and function, we selected four SUMF1 missense mutations for detailed analyses. These mutations had previously been identified in late infantile MSD patients in homozygous form or, in one heterozygous case (p.A177P), detectable FGE protein was likely to be expressed only from the missense allele.

Upon overexpression all alleles studied led to detectable FGE protein, which localized to the proper intracellular compartment, the ER. Due to overexpression, a fraction of wild type and mutant FGE was secreted to the cell culture medium. This indicates that at least some proportion of mutant protein passes the ER quality control system and after synthesis adopts correct folding and sufficient stability, so that the mutant FGEs in principle can fulfill their enzymatic function to a certain degree. We further show that all mutations, in particular p.A177P, p.W179S and p.R349W, dramatically impair the catalytic activity of mutant FGE without abolishing it completely. Clear differences were visible in intracellular protein retention. FGE secretion from overexpressing cells most likely results from saturation of the ER retention mechanism, which is still unknown. The changes observed under overexpressing conditions may therefore be less relevant in vivo, where effects on protein stability are certainly more important than on retention.

Table 3 summarizes the complete data set relevant for correlating the molecular and biochemical findings as well as the clinical phenotype with the four MSD causing mutations under study. The p.R349W allele was

10 Schlotawa et al.

predicted to destabilize FGE due to the loss of five side chain interactions [Dierks et al., 2005]. Indeed, this mutation leads to a highly instable FGE protein, which in addition has less than 1% of residual enzymatic activity as compared to wild type FGE. Protein expression in patient cells was severely reduced. Nevertheless, the mutant FGE could still be detected in its correct ER localization. Sulfatase activities in patient cells were drastically reduced. All MSD patients homozygous for p.R349W do have a severe form of the disease.

Table 3. Synopsis

aSee Dierks et al. [2005]. bSee Fig.1B (residual activity) and Fig.3 (stability and retention) for further detail; wt, wild type. cSee Fig.4 for further detail; ER, endoplasmic reticulum. dSee Table 1 for further detail.; LIM, late infantile mild; LIS, late infantile severe.

The p.W179S allele encodes an FGE protein with moderately reduced stability, but a very low residual

enzymatic activity of about 3%. In fact this mutation was predicted to affect substrate binding and catalytic activity [Dierks et al., 2005]. In patient fibroblasts mutant FGE was localized in the ER and the detectable protein signals were similar to wild type. Sulfatase activities were slightly diminished or even normal. This MSD patient exhibits a mild clinical phenotype.

Also the p.A177P mutation was predicted to affect catalytic functioning, as the mutated residue is located within the substrate binding groove [Dierks et al., 2005]. Consistently, we show here that this allele encodes a mutant protein with normal stability, but extremely low residual enzymatic activity (less than 1% of wild type). In patient fibroblasts mutant FGE was clearly detectable and localized to the ER. Sulfatase activities were reduced. The p.A177P MSD patient has a mild clinical phenotype.

As predicted, the p.A279V allele leads to instable FGE due to the insertion of a too large side chain [Dierks et al., 2005]. This mutant shows a high residual enzymatic activity of about 23%. In patient fibroblasts mutant FGE was correctly localized to the ER, but at a remarkably low level. Sulfatase activities were slightly reduced. Among all patients studied, this patient exhibits the mildest clinical phenotype with slow disease progression.

Taken together these results confirm the structure-based predictions of effects resulting from missense mutations in FGE. Mutations predicted to impair substrate binding and/or catalytic activity show a drastically reduced FGE activity, although the proteins are stable. Mutant FGE with slight (p.A279V) and severe conformational changes (p.R349W), predicted to lead to protein instability, are instable indeed. In addition, these stability mutants show either a high (p.A279V) or an extremely low residual activity (p.R349W), which can hardly be predicted merely from the localization of the mutated residue.

Using MSD patient fibroblasts we could confirm the results for mutant FGE stability that we had obtained in the HT-1080 expression system. High protein levels were seen for stable proteins and very low levels for instable proteins. Also in MSD fibroblasts mutant FGE is obviously localized in the ER. The residual sulfatase activities in these differentially affected fibroblasts reflect the biological functioning of FGE mutants. Severely reduced

Molecular Analysis of SUMF1 Mutations in Multiple Sulfatase Deficiency 11

sulfatase activities were measured in fibroblasts carrying the severe p.R349W mutation. All fibroblasts carrying the three milder mutations showed distinguishably higher sulfatase activities. Interestingly, some sulfatase activities were in the range of activities from wild type control fibroblasts. This phenomenon is known since the beginning of research on MSD [Chang et al., 1983]. As all sulfatase activities were measured from whole cell lysates, differences in sulfatase expression can not be excluded. Furthermore, differences in the stability of FGly-modified and unmodified sulfatases from wild type and MSD fibroblasts, respectively, are known, which probably affect steady-state sulfatase activities [Conary et al., 1988; Steckel et al., 1985]. It had also been described that MSD fibroblasts recover initially reduced sulfatase activity with ongoing time of cell culturing [Chang et al., 1983]. The molecular basis of this compensation phenomenon is unknown. This effect seems to be more visible in cells with milder mutations. All three cell lines with the severe p.R349W mutation showed extremely reduced sulfatase activities, although they were cultured as long as all other fibroblasts investigated. Notably, this observation has direct and important impact on the diagnosis of MSD which could fail if cells were grown too long before measuring sulfatase activities.

Our data lead us to conclude that there obviously is a relationship between genotype and phenotype in MSD depending on protein stability and residual activity of mutant FGE. The three most severely affected patients all carry the homozygous SUMF1 missense mutation p.R349W leading to both significant structural destabilization and drastically compromised catalytic function of FGE. The patients have a similar clinical presentation and can be clearly discriminated from the even more severe neonatal form. The three other MSD patients reveal comparable but milder clinical symptoms. They do have either stable FGE proteins (p.A177P, p.W179S) or high residual enzymatic activity (p.A279V). For establishing general genotype-phenotype correlation in MSD a more complete data set, describing the consequences of further SUMF1 missense and also nonsense mutations at both the molecular (FGE functionality) and the clinical level, is necessary. Such a data set could be very helpful for predicting disease severity and for precise genetic counseling of affected families. Until today we can only assume that SUMF1 null mutations, such as frameshift, truncation, splicing or translation initiation mutations, are associated with a complete loss of FGly generating activity and lead to the most severe neonatal phenotypes. So far, one MSD patient heterozygous for the mutation p.R327X, disrupting the active site of FGE, and a large deletion on the other allele (p.A149-A173del) supports this hypothesis. This patient showed neonatal disease onset with rapid progression (personal data). The majority of SUMF1 knock out mice, generated recently, die within the first month of age [Settembre et al., 2007]. In these mice sulfatase activities are completely absent.

Our data supporting a genotype-phenotype correlation in MSD are in contrast to results already published by Cosma and co-workers [2004]. They transiently co-expressed various sulfatases in Cos-7 cells together with MSD causing FGE mutants. The resulting levels of sulfatase activities showed no correlation with the severity of the clinical phenotype. It was concluded that FGE shows a differential affinity for the individual sulfatases, which varies for the different mutant FGE proteins. It was also postulated that other factors independent of FGE may influence the clinical phenotype of MSD. Thus, the question to which degree the molecular phenotype, studied at the level of the sulfatases or, as done here for the first time, at the level of FGE, correlates with the clinical phenotype, remains to be resolved. The surprisingly high and variable sulfatase activities measured in fibroblasts of more mildly affected patients (Table 2) suggest that it may be more reliable to study FGE functionality to define the molecular phenotype caused by SUMF1 mutations.

In conclusion, our functional analyses of the molecular defects caused by selected SUMF1 mutations in patients with late infantile MSD show that both, residual enzyme activity and protein stability of mutant FGE contribute to the clinical phenotype. This does not preclude that additional factors could influence FGE function and, in case of MSD, act as disease modifiers. Further detailed structural and functional investigations of FGE are required to understand the complex mechanisms of ER retention, turnover regulation, high-affinity binding of a large ensemble of newly synthesized sulfatase polypeptides, for which an unfolded state has to be maintained and, most important, FGly generation through a novel oxygenase mechanism involving a reducing cofactor [Dierks et al., 2005; Roeser et al., 2006]. This knowledge will certainly promote straightforward approaches for designing therapeutic strategies for this fatal disorder.

ACKNOWLEDGMENTS

We thank S.H. Green, K. Harzer, W.J. Kleijer, A. Kohlschütter, G.M. Mancini, and M.T. Zabot for providing clinical data. We are grateful to M. Balleininger, N. Eiselt, N. Hollstein, K. Neifer and T. Wilke for their technical assistance. Further, we thank S. L. Gande, I. Kalus, K. Kollmann, T. Lübke, M. Mariappan, A. Preusser-Kunze and

12 Schlotawa et al.

B. Schmidt for their support and helpful discussions. This work has been supported in part by the Research program, Faculty of Medicine, Georg August University of Göttingen (L.S.), by the Deutsche Forschungs-gemeinschaft, the Fonds der Chemischen Industrie and Shire Human Genetic Therapies, Inc. (Cambridge, MA).

REFERENCES

Burch M, Fensom AH, Jackson M, Pitts-Tucker T, Congdon PJ. 1986. Multiple sulphatase deficiency presenting at birth. Clin Genet 30: 409-415.

Chang PL, Rosa NE, Ballantyne SR, Davidson RG. 1983. Biochemical variability of arylsulphatases -A, -B and -C in cultured fibroblasts from patients with multiple sulphatase deficiency. J Inherit Metab Dis 6: 167-172.

Conary JT, Hasilik A, von Figura K. 1988. Synthesis and stability of steroid sulfatase in fibroblasts from multiple sulfatase deficiency. Biol Chem Hoppe-Seyler 369: 297-302.

Cosma MP, Pepe S, Annunziata I, Newbold RF, Grompe M, Parenti G, Ballabio A. 2003. The multiple sulfatase deficiency gene encodes an essential and limiting factor for the activity of sulfatases. Cell 113: 445-456.

Cosma MP, Pepe S, Parenti G, Settembre C, Annunziata I, Wade-Martins R, Di Domenico C, Di Natale P, Mankad A, Cox B et al. 2004. Molecular and functional analysis of SUMF1 mutations in multiple sulfatase deficiency. Hum Mutat 23: 576-581.

Diaz-Font A, Santamaria R, Cozar M, Blanco M, Chamoles N, Coll MJ, Chabas A, Vilageliu L, Grinberg D. 2005. Clinical, mutational characterization of three patients with multiple sulfatase deficiency: report of a new splicing mutation. Mol Genet Metab 86: 206-211.

Dierks T, Dickmanns A, Preusser-Kunze A, Schmidt B, Mariappan M, von Figura K, Ficner R, Rudolph MG. 2005. Molecular basis for multiple sulfatase deficiency and mechanism for formylglycine generation of the human formylglycine-generating enzyme. Cell 121: 541-552.

Dierks T, Lecca MR, Schlotterhose P, Schmidt B, von Figura K. 1999. Sequence determinants directing conversion of cysteine to formylglycine in eukaryotic sulfatases. EMBO J 18: 2084-2091.

Dierks T, Schmidt B, Borissenko LV, Peng J, Preusser A, Mariappan M, von Figura K. 2003. Multiple sulfatase deficiency is caused by mutations in the gene encoding the human Cα-formylglycine generating enzyme. Cell 113: 435-444.

Dierks T, Schmidt B, von Figura K. 1997. Conversion of cysteine to formylglycine: a protein modification in the endoplasmic reticulum. Proc Natl Acad Sci USA 94: 11963-11968.

Diez-Roux G, Ballabio A. 2005. Sulfatases and human disease. Annu Rev Genomics Hum Genet 6: 355-379.

Eto Y, Gomibuchi I, Umezawa F, Tsuda T. 1987. Pathochemistry, pathogenesis and enzyme replacement in multiple-sulfatase deficiency. Enzyme 38: 273-279.

Gieselmann V, Schmidt B, von Figura K. 1992. In vitro mutagenesis of potential N-glycosylation sites of arylsulfatase A. Effects on glycosylation, phosphorylation, and intracellular sorting. J Biol Chem 267: 13262-13266.

Hopwood JJ, Ballabio A. 2001. Multiple Sulfatase Deficiency and the nature of the sulfatase family. In: Scriver CR, Beaudet AL, Valle D, Sly WS, editors. The Metabolic and Molecular Bases of Inherited Disease. New York: McGraw-Hill. p 3725-3732.

Kepes JJ, Berry A 3rd, Zacharias DL. 1988. Multiple sulfatase deficiency: bridge between neuronal storage diseases and leukodystrophies. Pathology 20: 285-291.

Loffeld A, Gray RG, Green SH, Roper HP, Moss C. 2002. Mild ichthyosis in a 4-year-old boy with multiple sulphatase deficiency. Br J Dermatol 147: 353-355.

Mancini GM, van Diggelen OP, Huijmans JG, Stroink H, de Coo RF. 2001. Pitfalls in the diagnosis of multiple sulfatase deficiency. Neuropediatrics 32: 38-40.

Perlmutter-Cremer N, Libert J, Vamos E, Spehl M, Liebaers I. 1981. Unusual early manifestation of multiple sulfatase deficiency. Ann Radiol (Paris) 24: 43-48.

Preusser-Kunze A, Mariappan M, Schmidt B, Gande SL, Mutenda K, Wenzel D, von Figura K, Dierks T. 2005. Molecular characterization of the human Cα-formylglycine-generating enzyme. J Biol Chem 280: 14900-14910.

Molecular Analysis of SUMF1 Mutations in Multiple Sulfatase Deficiency 13

Roeser D, Preusser-Kunze A, Schmidt B, Gasow K, Wittmann JG, Dierks T, von Figura K, Rudolph MG. 2006. A general binding mechanism for all human sulfatases by the formylglycine-generating enzyme. Proc Natl Acad Sci USA 103: 81-86.

Rommerskirch W, von Figura K. 1992. Multiple sulfatase deficiency: catalytically inactive sulfatases are expressed from retrovirally introduced sulfatase cDNAs. Proc Natl Acad Sci USA 89: 2561-2565.

Sardiello M, Annunziata I, Roma G, Ballabio A. 2005. Sulfatases and sulfatase modifying factors: an exclusive and promiscuous relationship. Hum Mol Genet 14: 3203-3217.

Schmidt B, Selmer T, Ingendoh A, von Figura K. 1995. A novel amino acid modification in sulfatases that is defective in multiple sulfatase deficiency. Cell 82: 271-278.

Settembre C, Annunziata I, Spampanato C, Zarcone D, Cobellis G, Nusco E, Zito E, Tacchetti C, Cosma MP, Ballabio A. 2007. Systemic inflammation and neurodegeneration in a mouse model of multiple sulfatase deficiency. Proc Natl Acad Sci U S A 104: 4506-4511.

Steckel F, Hasilik A, von Figura K. 1985. Synthesis and stability of arylsulfatase A and B in fibroblasts from multiple sulfatase deficiency. Eur J Biochem 151: 141-145.

Vamos E, Liebaers I, Bousard N, Libert J, Perlmutter N. 1981. Multiple sulphatase deficiency with early onset. J Inherit Metab Dis 4: 103-104.

Voznyi YV, Keulemans JL, van Diggelen OP. 2001. A fluorimetric enzyme assay for the diagnosis of MPS II Hunter disease. J Inherit Metab Dis 24: 675-680.