a point mutation creating an extran-glycosylation site in fibrillin-1 results in neonatal marfan...

GENOMICS 36, 468 –475 (1996)ARTICLE NO. 0492

A Point Mutation Creating an Extra N-Glycosylation Site inFibrillin-1 Results in Neonatal Marfan Syndrome

LASSE LONNQVIST,*,1 LEENA KARTTUNEN,*,1 TERHI RANTAMAKI,* CAY KIELTY,†MICHAEL RAGHUNATH,‡ AND LEENA PELTONEN*,2

*Department of Human Molecular Genetics, National Public Health Institute, Mannerheimintie 166, FIN-00300 Helsinki, Finland;†Biochemistry Research Division, School of Biological Sciences, University of Manchester, Manchester, United Kingdom; and

‡Institute of Physiological Chemistry and Pathobiochemistry, University of Munster, Munster, Germany

Received May 1, 1996; accepted June 24, 1996

To date, over 50 family-specific FBN1 mutations re-Fibrillin-1 is a large cysteine-rich glycoprotein of the sulting in MFS have been reported (Dietz and Pyeritz,

10-nm microfibrils in the extracellular matrix. A spec- 1995). These mutations lead to a variety of conse-trum of mutations in the fibrillin-1 gene (FBN1) have quences including amino acid substitutions, deletions,been identified in patients with Marfan syndrome insertions, splicing errors, and premature termina-(MFS), and the majority of mutations resulting in the tions. The molecular mechanisms by which the muta-neonatal and often lethal form of MFS have been iden- tions cause the clinical disease are probably as diversetified in the restricted region of exons 24–32 of the as the spectrum of mutations and the variability ofFBN1 gene. Here we report a novel point mutation in the clinical MFS phenotype. Most mutations leading toexon 25 of the FBN1 gene in a patient with lethal MFS. amino acid substitutions probably disrupt the struc-The mutation resulted in a molecular defect rarely en- ture of calcium-binding epidermal growth factor (cb-countered in human diseases, the creation of an extra EGF)-like motifs, which in turn is thought to alter theconsensus sequence for N-glycosylation. Metabolic la- calcium-binding capacity of fibrillin-1. The cb-EGF-likebeling of the patient fibroblast culture and in vitro ex- motifs are repeated 43 times in the fibrillin-1 polypep-pression of the mutagenized cDNA construct suggest tide, and they are presumably necessary for cell–cell,that this novel N-glycosylation site is actually utilized. cell–matrix, and protein–protein interactions (Hand-Immunohistochemical and ultrastructural analyses of ford et al., 1991; Astermark et al., 1992; Hogg et al.,the fibroblast cultures of the patient show that this

1992). Another class of mutations, premature termina-excessive N-glycosylation severely affects microfibriltion codons, often cause a reduction in the steady stateformation in vitro; this finding emphasizes the impor-level of the mutant FBN1 transcript (Dietz et al., 1993;tance of correct posttranslational modifications of fi-Karttunen et al., 1996). A mutation resulting in intra-brillin molecules for correct aggregation into microfi-cellular retention, delay in secretion, and consequentbrillar structures. q 1996 Academic Press, Inc.posttranslational overmodification of profibrillin-1 hasalso been reported (Raghunath et al., 1995) as well asa mutation affecting the processing of profibrillin-1 intoINTRODUCTIONfibrillin-1 (Milewicz et al., 1995).

Here we report a novel molecular pathogenic mecha-Marfan syndrome (MFS) is a dominantly inheritednism in MFS that emphasizes the importance of themultiorgan disorder with typical manifestations in thenormal N-glycosylation of fibrillin-1 polypeptides. Askeletal, ocular, and cardiovascular systems (Pyeritz,novel point mutation in exon 25, resulting in a substitu-1993). MFS is relatively evenly distributed throughout tion of isoleucine by threonine, was identified in a pa-the world, and a majority of MFS cases are causedtient with neonatal MFS (nMFS) who died at the ageby mutations in the fibrillin-1 gene (FBN1) (Dietz and of 4 months. This amino acid change created a potential

Pyeritz, 1995). FBN1 codes for a 350-kDa precursor extra N-glycosylation site in the fibrillin-1 polypeptide,molecule, profibrillin-1, which is extracellularly and our experimental data provide evidence that thiscleaved to a mature, 320-kDa fibrillin-1 polypeptide novel N-glycosylation site is actually utilized. We sug-(Milewicz et al., 1995). Fibrillin-1 is the major compo- gest that excessive N-glycosylation of fibrillin-1 repre-nent of the 10-nm microfibrils of the extracellular ma- sents a rare molecular pathogenic mechanism in thistrix (Sakai et al., 1986). severe MFS case.

MATERIALS AND METHODS1 These authors contributed equally to this work.2 To whom correspondence should be addressed. Telephone: 358- Clinical summary. The patient was a newborn male, who had

congenital heart defects including a dilated aortic root and both mi-0-4744 393. Fax: 358-0-4744 480. E-mail: [email protected].

4680888-7543/96 $18.00Copyright q 1996 by Academic Press, Inc.All rights of reproduction in any form reserved.

AID Genom 4287 / 6r1c$$$321 08-18-96 22:30:03 gnma AP: Genomics

A MUTATION CREATING AN EXTRA N-GLYCOSYLATION SITE 469

tral and tricuspid regurgitation. In addition, he had severe arachno- twice with ice-cold Hank’s balanced salt solution and lysed as de-dactyly in the fingers and toes, extremely loose skin, and a bilateral scribed earlier (Raghunath et al., 1995). The total proteins of the cellinguinal hernia. Additional features included sustained fractures of lysates were analyzed with 8% SDS–PAGE and fluorography.the clavicle and humerus. The diagnosis of MFS was established at Immunoprecipitation and endoglycosidase H and N-glycosidase Fbirth. The patient had major respiratory problems and died at the digestions. Radiolabeled neonatal polypeptides from the cell cultureage of 4 months. Autopsy was not authorized so the actual immediate medium were immunoprecipitated using Protein A–Sepharosecause of death is not known. Neither of the parents displayed any (Sigma) as described earlier (Paunio et al., 1994) with polyclonalsigns of MFS. antibodies raised in rabbits against the recombinant neonatal poly-

DNA extraction, PCR, single-strand conformation polymorphism peptide (T. Rantamaki, Natl. Publ. Health Inst., Helsinki, Finland,(SSCP) analysis, and sequencing. DNA was isolated from periph- pers. comm.). Washed immunoprecipitates were solubilized and di-eral blood as described earlier (Vandenplas et al., 1984), with minor gested with 0.1 U endoglycosidase H (Endo H) or 0.08 U N-glycosi-modifications. PCR-amplification of exon 25 of the FBN1 gene was dase F (New England Biolabs) essentially as described earlier (Tikka-carried out using the primers 5*-bio-GGC CTC TGC CAC AGA GGT nen et al., 1995). Prior to N-glycosidase F digestion, N-octylglucosideCA-3* and 5*-CAG CCT CTG CAC CCA CGG CA-3*. Solid-phase was added to 1% (w/v) to overcome the inhibitory effect of SDS onsequencing of the amplification product was carried out using the N-glycosidase F. Digestions were performed overnight and analyzedprimer 5*-CAG CCT CTG CAC CCA CGG CA-3 * as described earlier with SDS–PAGE and fluorography.(Syvanen et al., 1989). Screening of the genomic DNA samples of 49 Metabolic labeling of the fibroblast cultures. Three days prior tounrelated MFS patients was carried out similarly. The sequencing metabolic labeling, 250,000 dermal fibroblasts from the patient andreactions were analyzed by denaturing polyacrylamide gel electro- an age-matched control individual were plated in 35-mm dishesphoresis. (Nunc) and maintained in DMEM supplemented with 10% FCS and

Screening of the coding region of the FBN1 gene for other possible penicillin–streptomycin. On Day 4, the cells were starved in DMEMmutations in the patient was performed by SSCP analysis as de- in the absence of FCS, methionine, and cysteine for 1 h, after whichscribed earlier (Kainulainen et al., 1992, 1994). the cells were labeled for 1 h with a mixture of [35S]methionine and

Screening of the genomic DNA samples of the parents of the pro- [35S]cysteine (SJ1015 and SJ15232, respectively, Amersham) con-band, as well as of a control pool consisting of genomic DNA samples taining 50 mCi of each label in 500 ml of Cys- and Met-free DMEMof 180 control individuals, was carried out using the solid-phase mini- without FCS, and chased for 1, 2, and 4 h. The cells were harvestedsequencing method (Syvanen et al., 1993). Exon 25 was amplified and the total proteins were visualized with 4%/7% step SDS–PAGEusing the primers described above, and the sequence of a detection and fluorography as described earlier (Raghunath et al., 1995).primer in the solid-phase minisequencing reaction, designed to an-

Densitometrical analyses. The fragment representing vimentin,neal immediately upstream of nucleotide T-3143, was 5*-CAC GGCan intracellular protein, was selected as an indicator of the totalAAG TGC AGA AAC ACC A-3*.protein mass. Autoradiographs representing cell lysate samples of

The expression construct. The 69-bp signal sequence of a gene both the patient and the control were densitometrically scanned forfor a lysosomal enzyme, aspartylglucosaminidase (AGA), was PCR- the integrated optical density (IOD) of the vimentin fragment. Theamplified from the AGA cDNA clone (Ikonen et al., 1991) with prim- amounts of the total protein loaded on the second SDS–PAGE wereers flanking the signal sequence and containing recognition se- calibrated to be identical in both samples. The medium samplesquences for HindIII and KpnI restriction enzymes. The PCR product loaded on SDS–PAGE were also calibrated according to the amountwas cloned into SV-poly expression vector (Stacey and Schnieke,

of intracellular protein mass, for both the control and the patient.1990).SDS–PAGE and fluorography were performed as described above

Bluescript II SK(/) vector containing the FBN1 cDNA, encoding and the IOD values of the fragment representing profibrillin-1 in thenucleotides 2694–8787 (kindly provided by Dr. Dieter Reinhardt,

cell lysate samples, and the fragments representing profibrillin-1Portland, Oregon), was digested with FokI (Promega), and the largest and fibrillin-1 in the medium samples, were densitometricallyof the resulting fragments—1889 bp, consisting of nucleotides 2843– scanned using a Millipore BioImage scanner (Millipore).4732 and corresponding to exons 24–37 of FBN1 cDNA—was puri-

Immunofluorescence experiments. Dermal fibroblasts (106) fromfied from an agarose gel. The ends of the FBN1 fragment were filledthe patient and an age-matched control individual were plated inwith Klenow fragment (Promega). The resulting blunt-ended cDNA35-mm dishes and cultured in DMEM, supplemented as describedfragment was ligated to the SV-poly expression vector containingearlier, for 72 h. To inhibit N-glycosylation, tunicamysin was addedthe AGA signal sequence and digested with EcoRV (Promega). Theto the cell cultures at a final concentration of 1 mg/ml. The resultingresulting clones were sequenced and the construct in the correcthyperconfluent cell layers were processed and immunostained withreading frame was selected for in vitro expression studies.anti-fibrillin-1 mAb F2 (kindly provided by Dr. L. Sakai, Portland,The T3143C mutation was introduced into the wildtype constructOregon) or with a commercially available polyclonal anti-fibronectinusing an in vitro mutagenesis kit in accordance with the manufactur-antibody (DAKO) as described earlier (Hollister et al., 1990).er’s protocol (Transformer Site-Directed Mutagenesis Kit, PT1130-

1, Clontech) with an antisense primer to introduce the mutation Ultrastructural analyses. Fibroblasts of the patient and the con-(CTT AAA GCT GCC AGT GGT GTT TCT G) and an antisense selec- trol individual were grown in parallel. The cells were plated at con-tion primer to delete a unique SacI restriction site from the construct fluence and maintained for 3 weeks prior to analysis. For microfibril(CGA ATT CCC GGG ACC GCT CGA T). extractions the cell layers were digested with 0.1 mg bacterial colla-

genase per milliliter (type 1A, Sigma) for 2 h at 207C. ImmunogoldTransfection of COS-1 cells, pulse-labeling, and immunoprecipita-RSEM for immunolocalization of fibrillin-1 polymers was performedtion of the synthesized polypeptides. One day prior to transfection,as described earlier (Sheenan et al., 1987) with minor modifications.400,000 COS-1 cells were plated in 35-mm dishes (Nunc) and main-Briefly, void volume fractions were adsorbed onto carbon-coatedtained in Dulbecco’s modified Eagle’s medium (DMEM) supple-nickel grids and then incubated sequentially by ‘‘hanging drop’’ asmented with 10% fetal calf serum (FCS) and penicillin –streptomy-follows: 31 5-min washes with 5 mM magnesium acetate containingcin. The cells were transfected with 5 mg of the wildtype or the mutant1% Tween 20 (MAT) and polyclonal anti-fibrillin-1 antiserum 5507neonatal construct using the DEAE-dextran transfection methoddiluted 1:1000 in MAT for 2 h at room temperature, 31 washes in(Luthman and Magnuson, 1983). On the third day after transfection,MAT and protein G–gold conjugate (Sigma) diluted 1:100 in MAT,the cells were either harvested or pulse-labeled. To inhibit N-glyco-31 5-min washes in MAT, and then one wash in 5 mM magnesiumsylation, the cells were treated with 5 mg/ml tunicamycin and thereaf-acetate. Finally, the grids were dipped in ethanol, air dried, and thenter pulse-labeled. The transfected cells were starved for 1 h in Cys-shadowed at an angle of 57 using a tungsten–platinum filament. Thefree DMEM without FCS. The cells were labeled for 1 h with [35S]-shadowed samples were visualized using a JEOL 1200EX electroncysteine (SJ15232, Amersham), 100 mCi in 500 ml Cys-free DMEM

without FCS. The medium was removed, and the cells were washed microscope at 120 kV.


LONNQVIST ET AL.470

times as strong as the corresponding fragment in thecontrol cells. The ratio of the intracellular profibrillin-1 and the fibrillin-1 present in the culture medium wasalso determined. The fraction present in the mediumconsisted of both profibrillin-1 and fibrillin-1, andtherefore the values of the integrated optical densitiesof these two molecular forms were combined. However,no significant difference in the cell/medium ratio be-tween the control (0.52) and the patient (0.57) samplecould be detected.

Immunofluorescence Experiments

To analyze the consequences of the mutation for theFIG. 1. Detection of the mutation at the DNA level. DNA se- microfibril network in the extracellular matrix, the pa-

quence from a control individual (C) and the patient (P). Arrow showstient and the control fibroblast cultures were immuno-the T-to-C transition at nucleotide 3143. The codon ATT coding forhistochemically stained with an anti-fibrillin-1 anti-isoleucine is changed to ACT, which codes for threonine. The mutant

nucleotide is marked with an asterisk. body. The immunofluorescence pattern in the patientcell culture showed only some unorganized aggregatesand the hyperconfluent cell layer underneath was

RESULTS clearly visible (Fig. 3A). No fibrillar meshwork, typi-cally observed in the control culture (Fig. 3C), was ob-

Mutation Identification served in the patient cell culture. Interestingly, whenthe fibroblast culture of the patient was treated withThe phenotype of the patient is summarized undertunicamycin, which inhibits N-glycosylation, more im-Materials and Methods. The systematic sequence anal-munoreactive material accumulated than in the un-ysis of the ‘‘neonatal region’’—exons 24–32—revealedtreated patient cell culture (Fig. 3B). In contrast to this,a T-to-C transition at nucleotide 3143 (T3143C) in exonthe immunofluorescence staining of the tunicamycin-25 of the FBN1 gene of the nMFS patient (Fig. 1). Thistreated control cells demonstrated impaired formationwould result in an amino acid change from isoleucineof the fibrillar network compared to the staining of theto threonine at position 1048 (I1048T) of the fibrillin-untreated control cells (Fig. 3D). In both the presence1 polypeptide. The substitution creates a novel poten-and the absence of tunicamycin, the immunofluores-tial N-glycosylation site at Asn-1046, with the localcence patterns of the patient and the control fibroblastsequence being changed from Asn-Thr-Ile to Asn-Thr-cultures stained with anti-fibronectin antibodies re-Thr, which is in agreement with the consensus se-vealed an identical pattern, which was reduced in thequence for N-glycosylation, Asn-Xaa-Thr/Ser (Xaa de-presence of tunicamycin (Fig. 4). This would suggestnotes any amino acid except proline). The mutation wasthat the observed ability of tunicamycin to increase thenot found in the genomic DNA samples of the patient’samount of immunoreactive material in the patient cellparents, in 49 unrelated MFS patients, or in the poolculture carried some specificity for fibrillin-1.of 180 control individuals, showing that the mutation

did not represent a common polymorphism. After iden-tification of the mutation, the coding region of theFBN1 gene of the patient was screened for other possi-ble mutations by SSCP analysis. This analysis revealedno other mobility shifts, suggesting that no other muta-tions were present in the patient sample.

Metabolic Labeling of the Fibroblast Cultures

To obtain evidence as to whether the novel potentialN-glycosylation site is actually utilized, we analyzed

FIG. 2. Pulse –chase analysis of the fibroblast cultures. Cell ly-fibroblast cultures from the patient and from an age-sates analyzed by 4%/7% step SDS–PAGE (Raghunath et al., 1995)matched control individual in parallel. The total meta-from the metabolically labeled fibroblast cultures of the patient (P)bolically labeled proteins derived from lysed fibroblast and the control individual (C). Cell lysates are shown at the 2-h

cultures were analyzed by SDS–PAGE and fluorogra- time-point. An arrow indicates the position of profibrillin-1, the intra-phy. After a 2-h chase, in the patient sample, a popula- cellular precursor molecule, approximately 350 kDa, which is enzy-

matically cleaved to fibrillin-1 polypeptide, approximately 320 kDa.tion of profibrillin-1 molecules migrating more slowlyAt the 2-h chase time-point, the more slowly migrating profibrillin-than the corresponding profibrillin-1 in the control1 is shown in the sample of the patient. The anonymous intracellularsample was observed (Fig. 2). The integrated optical ‘‘protein x,’’ which is the largest detectable polypeptide detectable in

density of the radiolabeled fragment representing pro- this gel system and which migrates just above profibrillin-1, is alsomarked with an arrow.fibrillin-1 in the patient cells also revealed a signal 2.0



FIG. 3. Immunofluorescence staining of hyperconfluent patient and control fibroblast cultures with a monoclonal anti-fibrillin-1 antibodyshowing the patient culture to be almost completely devoid of fibrillin-containing microfibrillar structures, in contrast to the control culture.After tunicamycin treatment, slightly more immunoreactive material can be seen in the patient culture, but the organization of fibrillarstructures after tunicamycin treatment is severely impaired, which can also be seen in the control culture. (A) The patient fibroblast culture.(B) The patient fibroblast culture treated with tunicamycin. (C) The control fibroblast culture. (D) The control fibroblast culture treatedwith tunicamycin.

FIG. 4. Immunofluorescence staining of hyperconfluent patient and control fibroblast cultures with polyclonal anti-fibronectin antibodiesshowing equal staining of both the patient and the control fibroblast cultures and impaired fibronectin staining after tunicamycin treatment.(A) The patient fibroblast culture. (B) The patient fibroblast culture treated with tunicamycin. (C) The control fibroblast culture. (D) Thecontrol fibroblast culture treated with tunicamycin.

the wildtype and the mutant polypeptide on SDS–Ultrastructural AnalysesPAGE could be observed. In addition, the polypeptides

Rotary shadowing electron microscopy of the depos- migrated faster than the corresponding wildtype poly-ited matrix, stained with immunogold, demonstrated peptide synthesized in the absence of tunicamycin.the absence of any stable fibrillin-1-containing micro-fibrillar assemblies in the patient fibroblast culture(Fig. 5A). Only globular structures were observed.These globules most probably correspond to the ‘‘beads’’of the characteristic beads-on-a-string-like appearanceof assembled 10-nm microfibrils seen in the control cul-ture (Fig. 5B).

In Vitro Expression of the Neonatal cDNA Construct

Fibrillin-1 is a large protein with 15 putative N-gly-cosylation sites. Since it is fairly difficult to analyze asmall migration difference caused by a putative extrasugar chain in the complete 350-kDa fibrillin-1 propep-tide, we constructed a ‘‘neonatal expression construct’’containing the coding region for exons 24–37 of theFBN1 gene. The neonatal construct encodes a polypep-tide of 666 amino acids with four putative N-glycosyla-tion sites. COS-1 cells were transfected with the wild-type and the in vitro mutagenized (I1048T) neonatalconstruct. SDS–PAGE analysis of the pulse-labeled

FIG. 5. Rotary shadowing electron microscopy. 10-nm microfi-cell lysate samples revealed that the synthesized mu-brils isolated from postconfluent fibroblast layers. (A) The patienttant fibrillin-1 polypeptide migrated substantiallysample. (B) The control sample. The patient sample consists of onlyslower than the wildtype polypeptide (Fig. 6). When the globular structures that most probably correspond to the ‘‘beads’’ of

N-glycosylation in the transfected cells was inhibited the-beads-on-a-string-like microfibrils, which are typically seen inthe control sample.using tunicamycin, no migration difference between


LONNQVIST ET AL.472

tient described in this paper displayed typical nMFSand died at the age of 4 months.

Not only clinically but also on the basis of its molecu-lar background, nMFS represents an interesting en-tity. In the vast majority of reported nMFS cases, thecausative mutation has been detected in exons 24–32of the FBN1 gene (Kainulainen et al., 1994; Milewiczet al., 1994; Wang et al., 1995). It seems clear thatthis ‘‘neonatal region’’ of the fibrillin-1 polypeptide isof special importance for the integrity and stability of

FIG. 6. In vitro expression of the neonatal cDNA construct. 8% fibrillin-1-containing microfibrils. This assumption isSDS–PAGE from the total proteins of cell lysates of the metabolically supported not only by the severity of the clinical pheno-labeled COS-1 cells that were transiently transfected with the neona-

type of nMFS patients but also by the consistent find-tal cDNA constructs. Wt, polypeptide encoded by the neonatal con-ings reported in immunohistochemical and ultrastruc-struct carrying the wildtype neonatal construct; Mut, polypeptide

encoded by the neonatal construct with the I1048T mutation; Cos, tural analyses of nMFS patients’ cell cultures. In im-the COS-1 cell background.0, Cells without addition of tunicamycin; munohistochemical studies, nMFS fibroblast cultures/, cells treated with tunicamycin. Arrows indicate the position of

have been shown to be completely devoid of immunore-the polypeptide translated from the construct. Migration of the mo-active material when stained with anti-fibrillin-1 anti-lecular weight standards is marked with bars. The migration differ-

ence between the polypeptides is clearly seen when no tunicamycin bodies (Superti-Furga et al., 1992; Raghunath et al.,is added to the cells. After treatment with tunicamycin, no migration 1993), and in rotary shadowing electron microscopydifference can be observed. analyses, nMFS cell cultures have typically displayed

no identifiable 10-nm-diameter microfibrils at all (C.Kielty, Univ. of Manchester, Manchester, UK, pers.This suggests that the polypeptide translated from thecomm.). In our patient too, both immunofluorescenceneonatal construct is actually N-glycosylated and thatand rotary shadowing electron microscopy analysis ofexcessive N-glycosylation was responsible for the ob-the fibroblast culture showed a typical staining pat-served mobility difference in the mutant polypeptide.tern, with an almost complete absence of microfibrillarTo analyze the N-linked oligosaccharides in the poly-structures.peptide translated from the neonatal construct further,

Fibrillin-1 is a major component of 10-nm microfi-the expressed wildtype and mutant polypeptides fromthe medium were digested with endoglycosidase H and brils that are widely distributed in the human bodyN-glycosidase F (Maley et al., 1989). On the SDS– and are believed to function as a scaffold for the deposi-PAGE, the undigested wildtype polypeptide migrated tion and formation of elastic fibers (Sakai et al., 1986).as an approximately 120-kDa fragment according to Fibrillin-1 consists of several different motifs of whichthe molecular weight marker and the corresponding the six-cysteine, cb-EFG-like domains are the mostsize of the I1048T mutant polypeptide was about 122 abundant (Corson et al., 1993; Pereira et al., 1993). ThekDa. Digestion with both endoglycosidase H and N- specific disulfide bridge formation and a consensus siteglycosidase F abolished this size difference, with the for b-hydroxylation of cb-EGF-like domains are in-size of both the wildtype and the mutant polypeptides volved in the calcium binding of these motifs, which isbeing 100 kDa (Fig. 7). These data indicate that the thought to be important for the stability of fibrillin-1size difference between the wildtype and the mutant (Corson et al., 1993). The other motifs of fibrillin-1 in-polypeptide is due to an extra oligosaccharide residue,and sensitivity to endoglycosidase H suggests that allthe N-linked oligosaccharides are of a high-mannosetype.

DISCUSSION

A variety of clinical phenotypes of MFS have beenshown to result from mutations in the FBN1 gene(Dietz and Pyeritz, 1995). The spectrum ranges from

FIG. 7. Endoglycosidase H and N-glycosidase F digestions of theectopia lentis and marfanoid skeletal features, which immunoprecipitated polypeptides encoded by the wildtype and therepresent the ‘‘mildest’’ forms of FBN1-related disor- mutant neonatal constructs from the medium of transiently trans-ders, to rapidly progressive and often neonatally lethal fected COS-1 cells analyzed by 8% SDS–PAGE. Wt, polypeptide en-

coded by the neonatal construct carrying the wildtype neonatal con-Marfan syndrome, which is the most severe clinicalstruct; Mut, polypeptide encoded by the neonatal construct with theexpression of the MFS phenotype continuum (Dietz andI1048T mutation; Cos, the COS-1 cell background. 0, UndigestedPyeritz, 1995). nMFS patients typically suffer from a samples; H, endoglycosidase H digestion; F, N glycosidase F diges-

variety of severe cardiovascular problems that usually tion. Arrows indicate the size of the polypeptides before digestion(top two arrows) and after digestion (bottom arrow).result in death during the first months of life. The pa-



clude domains resembling the one identified in trans- type of oligosaccharide was analyzed by digesting theexpressed polypeptides with endoglycosidase H and N-forming growth factor b-binding protein (TGF-b-bp-

like), a region rich in proline residues, and unique glycosidase F (Maley et al., 1989). All the oligosaccha-rides, including the one attached to the newly formedamino- and carboxy-terminal regions (Pereira et al.,

1993). Fibrillin-1 also contains 15 putative sites for N- N-glycosylation site, were endoglycosidase H-sensitive,indicating that all the sugars are of a high-mannoseand several putative sites for O-glycosylation, but how

many of these are actually used is not known (Glanville type. The neonatal fragment represents a truncatedfibrillin-1 polypeptide, and its posttranslational modi-et al., 1994). The specific function of the different re-

gions in fibrillin-1 fiber formation or in the interactions fications may differ from those of the full-length poly-peptide. However, the expressed polypeptide containsof fibrillin with other ECM proteins is not yet under-

stood. the longest uninterrupted stretch of cb-EGF-like motifsin the fibrillin-1 polypeptide, 12 altogether, and thisIn this report, we describe an FBN1 mutation re-

sulting in the formation of an extra N-glycosylation rigid domain with multiple disulfide bridges most prob-ably folds independently of the surrounding structuresconsensus sequence and provide evidence that this site

is actually utilized. The patient had a mutation re- (Cooke et al., 1987). This truncated polypeptide is prob-ably not treated differently from the corresponding re-sulting in an amino acid change from isoleucine to thre-

onine, in exon 25. Computer-based analysis of this par- gion in the full-length fibrillin-1 polypeptide by theposttranslational modification machinery.ticular cb-EGF-like motif demonstrated no radical

changes in the secondary structure, and the isoelectric Most of the secretory proteins undergo N-glycosyla-tion. The sugar chains are important for the correctpoint of the mutant cb-EGF like motif also remained

unchanged. This was largely expected, since the muta- folding and stability of the polypeptide, and it has beenshown that oligosaccharides also participate in pro-tion does not disrupt the disulfide bridges of the motif

nor does this isoleucine residue represent a conserved tein–protein interactions (Hart, 1992). Mutations af-fecting the potential N-glycosylation sites in proteinsamino acid among the cb-EGF-like motifs of fibrillin-1.

The creation of an additional potential N-glycosylation have so far been reported in only a few other humandiseases. Mutations leading to the disruption of an ex-site in the fibrillin-1 polypeptide by this point mutation

appears to be more important for the pathogenesis of isting potential N-glycosylation site have been re-ported, for example, in a sectorial form of autosomalthe disease. The electrophoretic mobility difference of

the profibrillin-1 polypeptides between the patient and dominant retinitis pigmentosa (Sullivan et al., 1993),in a saposin B deficiency (Kretz et al., 1990), and in athe control sample was demonstrated in the cell lysate

samples. Interestingly, the immunoreactive material patient suffering from glycogen storage disease type II(Hermans et al., 1993). At least one disease-causingin the patient cell cultures analyzed by immunofluo-

rescence staining with an anti-fibrillin antibody mutation leading to the formation of a potential extraN-glycosylation site has earlier been reported. In aseemed to increase when the cells were treated with a

trace amount of tunicamycin, which completely inhib- patient suffering from dysfibrinogenemia associatedwith posttraumatic bleeding, a methionine-to-threo-its the N-glycosylation of all the synthesized proteins

of the cell. This observation could be due to the specific- nine substitution at position 310 of the fibrinogen poly-peptide resulted in glycosylation of asparagine-308 dueity of the antibody, which is not capable of recognizing

over-N-glycosylated fibrillin-1 but more easily recog- to the newly formed N-glycosylation consensus se-quence (Yamazumi et al., 1989). The bulky extra oligo-nizes the underglycosylated form. This finding would

also favor the hypothesis that mutant fibrillin mole- saccharide structure was thought to impair the poly-merization of fibrinogen molecules severely. A similarcules are secreted from the cell and severely impair the

fibrillin fiber formation process. kind of effect, hampering fiber formation, could verywell be possible in the case of an excessively glycosyl-Since fibrillin-1 is such a large, 320-kDa protein, a

neonatal expression construct consisting of exons 24– ated fibrillin-1 molecule. Excessive glycosylation mayalso result in retarded secretion of the mutant fibrillin-37 of the FBN1 gene was designed to facilitate more

reliable analysis of migration differences on SDS– 1 polypeptide. We have shown here that the patientcell cultures had more intracellular profibrillin-1 thanPAGE. In vitro expression of the mutant neonatal con-

struct displayed a significant migration difference be- the control cell cultures, but the ratio between intracel-lular and extracellular fibrillin-1 polypeptides was sim-tween the wildtype and the mutant polypeptide. Inhibi-

tion of N-glycosylation by tunicamycin resulted in poly- ilar in both samples. However, this does not excludethe possibility of intracellular retention, since the ex-peptides of equal size from both the mutant and the

wildtype neonatal samples, strongly suggesting that tracellular proportion of the free fibrillin-1 polypep-tides may well be increased due to impaired incorpora-the new putative N-glycosylation site may be used. In

addition, the polypeptides migrated faster than the tion of the mutant, excessively glycosylated fibrillin-1 polypeptides into microfibrils. Severe impairment ofwildtype polypeptide from the cells grown without tuni-

camycin, indicating that at least some of the four puta- fiber formation was actually shown in immunofluores-cence analyses of the patient fibroblast cultures.tive N-glycosylation sites normally present in the wild-

type neonatal construct are used. Furthermore, the In conclusion, our data suggest a new molecular


LONNQVIST ET AL.474

precursor epidermal growth factor-like domains of fibrillin-1, thepathogenic mechanism in MFS that emphasizes theMarfan gene protein. J. Biol. Chem. 269: 26630–26634.importance of the correct glycosylation of the fibrillin-1

Handford, P. A., Mayhew, M., Baron, M., Winship, P. R., Campbell,polypeptide. Asparagine 1046, to which the additionalI. D., and Brownlee, G. G. (1991). Key residues involved in calcium-sugar chain is linked, is normally a consensus site for binding motifs in EGF-like domains. Nature 351: 164 –167.

b-hydroxylation in cb-EGF-like motifs (Cooke et al., Hart, G. W. (1992). Glycosylation. Curr. Opin. Cell Biol. 4: 1017–1987). The extra sugar chain linked to the polypeptide 1023.probably inhibits the b-hydroxylation of the asparagine Hermans, M. M., de Graaff, E., Kroos, M. A., Wisselaar, H. A., Wil-residue and could also locally disturb the folding of this lemsen, R., Oostra, B. A., and Reuser, A. J. (1993). The conserva-

tive substitution Asp-645 r Glu in lysosomal alpha-glucosidaseparticular cb-EGF-like motif. These events, in turn,affects transport and phosphorylation of the enzyme in an adultmost likely cause disturbance in the calcium-bindingpatient with glycogen-storage disease type II. Biochem. J. 289:ability of the mutant fibrillin-1 polypeptide and result 687–693.

in impaired formation or decreased stability of the fi- Hogg, P. J., Ohlin, A-K., and Stenflo, J. (1992). Identification of struc-brillin-1-containing microfibrils in the dominant nega- tural domains in protein c involved in its interaction with throm-tive fashion (Dietz and Pyeritz, 1995). It seems clear bin-thrombomodulin on the surface of endothelial cells. J. Biol.

Chem. 267: 703–706.that FBN1 mutations result in a wide variety of defectsHollister, D. W., Godfrey, M., Sakai, L. Y., and Pyeritz, R. E. (1990).in the posttranslational modifications, not only by dis-

Immunohistologic abnormalities of the microfibrillar-fiber systemturbing the formation of disulfide bridges but also byin the Marfan syndrome. N. Engl. J. Med. 323: 152–159.affecting the glycosylation and proteolytic processing

Ikonen, E., Baumann, M., Gron, K., Syvanen, A-C., Enomaa, N.,of fibrillin-1 polypeptides. Detailed analyses of the con-Halila, R., Aula, P., and Peltonen, L. (1991). Aspartylglucosaminu-

sequences of different FBN1 mutations increase our ria: cDNA encoding human aspartylglucosaminidase and the mis-knowledge of these modifications and also of the normal sense mutation causing the disease. EMBO J. 10: 51–58.function of fibrillin-1. Unraveling these events is essen- Kainulainen, K., Karttunen, L., Puhakka, L., Sakai, L., and Pelto-

nen, L. (1994). Mutations in the fibrillin gene responsible for domi-tial for further understanding of the molecular patho-nant ectopia lentis and neonatal Marfan syndrome. Nature Genet.genesis of individual FBN1 mutations and the re-6: 64–69.sulting high variability in MFS phenotypes.

Kainulainen, K., Sakai, L., Child, A., Pope, M., Puhakka, L., Ryha-nen, L., Palotie, A., Kaitila, I., and Peltonen, L. (1992). Two muta-

ACKNOWLEDGMENTS tions in Marfan syndrome resulting in truncated fibrillin polypep-tides. Proc. Natl. Acad. Sci. USA 89: 5017–5921.

We thank Drs. E. Hatchwell, K. Temple (Southampton, UK), and Karttunen, L., Lonnqvist, L., Godfrey, M., Peltonen, L., and Syvanen,H. Penford (Salisbury, UK) for the blood and fibroblast samples of A-C. (1996). An accurate method for comparing transcript levelsthe patient and his parents. We thank Dr. Lynn Sakai (Portland, of two alleles or highly homologous genes: Application to fibrillinOregon) for providing us with the anti-fibrillin-1 antibody F2 and transcripts in Marfan patients fibroblasts. Genome Res. 6: 392 –Dr. Dieter Reinhardt (Portland, Oregon) for providing us with the 403.FBN1 cDNA. This study was financially supported by the Academy

Kretz, K. A., Carson, G. S., Morimoto, S., Kishimoto, Y., Fluharty,of Finland, the Maud Kuistila Foundation, the Finnish FoundationA. L., and O’Brien, J. S. (1990). Characterization of a mutation infor Cardiovascular Research, the Aarno and Aili Turunen Founda-a family with saposin B deficiency: A glycosylation site defect. Proc.tion, the Finnish Medical Society Duodecim, the Orion-Farmos Foun-Natl. Acad. Sci. USA 87: 2541–2544.dation, the Medical Research Council (UK), and the Swiss National

Foundation. Luthman, H., and Magnusson, G. (1983). High efficiency polyomaDNA transfection of chloroquine treated cells. Nucleic Acids Res.11: 1295–1307.REFERENCES

Maley, F., Trimble, R. B., Tarentino, A. L., and Plummer, T. H.(1989). Characterization of glycoproteins and their associated oli-Astermark, J., Hogg, P. J., Bjork, I., and Stenflo, J. (1992). Effectsgosaccharides through the use of endoglycosidases. Ann. Biochem.of g-carboxyglutamic acid and epidermal growth factor-like mod-180: 195–204.ules of factor IX on factor X activation. J. Biol. Chem. 267: 3249–

3256. Milewicz, D. M., and Duvic, M. (1994). Severe neonatal Marfan syn-drome resulting from a de novo 3-bp insertion into the fibrillinCooke, R. M., Wilkinson, A. J., Baron, M., Pastore, A., Tappin, A. J.,gene on chromosome 15. Am. J. Hum. Genet. 54: 447–453.Campbell, I. D., Gregory, H., and Sheard, B. (1987). The solution

structure of human epidermal growth factor. Nature 327: 339– Milewicz, D. M., Grossfield, J., Cao, S-H., Kielty, C., Covitz, W.,341. and Jewett, W. (1995). A mutation in FBN1 disrupts profibrillin

processing and results in isolated skeletal features of the MarfanCorson, G. M., Chalberg, S. C., Dietz, H. C., Chardonneau, N. L.,syndrome. J. Clin. Invest. 95: 2373–2378.and Sakai, L. Y. (1993). Fibrillin binds calcium and is coded by

cDNAs that reveal a multidomain structure and alternatively Paunio, T., Kangas, H., Kalkkinen, N., Haltia, M., Palo, J., and Pelto-spliced exons at the 5* end. Genomics 17: 476–484. nen, L. (1994). Toward understanding the pathogenic mechanisms

in gelsolin-related amyloidosis: In vitro expression reveals an ab-Dietz, H. C., and Pyeritz, R. E. (1995). Mutations in the humannormal gelsolin fragment. Hum. Mol. Genet. 3: 2223–2229.gene for fibrillin-1 (FBN1) in the Marfan syndrome and related

disorders. Hum. Mol. Genet. 4: 1799–1809. Pereira, L., D’Alessio, M., Ramirez, F., Lynch, J. R., Sykes, B., Pangil-inan, T., and Bonadio, J. (1993). Genomic organization of the se-Dietz, H. C., McIntosh, I., Sakai, L. Y., Corson, G. M., Chalberg,quence coding for fibrillin, the defective gene product in MarfanS. C., Pyeritz, R. E., and Francomano, C. A. (1993). Four novelsyndrome. Hum. Mol. Genet. 2: 961–968.FBN1 mutations: Significance for mutant transcript level and

EGF-like domain calcium binding in the pathogenesis of Marfan Pyeritz, R. E. (1993). Marfan syndrome. In ‘‘Connective Tissue andsyndrome. Genomics 17: 468 –475. Its Heritable Disorders: Molecular, Genetic and Medical Aspects’’

(P. M. Royce and B. Steinmann, Eds.), pp. 437–468, Wiley–Liss,Glanville, R. W., Qian, R-Q., McClure, D. W., and Maslen, C. L.(1994). Calcium binding, hydroxylation, and glycosylation of the New York.



Raghunath, M., Kielty, C. M., and Steinmann, B. (1995). Truncated Syvanen, A-C., Aalto-Setala, K., Kontula, K., and Soderlund, H.(1989). Direct sequencing of affinity-captured amplified humanprofibrillin of a Marfan patient is of apparent similar size as fibril-

lin: Intracellular retention leads to over N-glycosylation. J. Mol. DNA: Application to the detection of apolipoprotein E polymor-phism. FEBS Lett. 258: 71–74.Biol. 248: 901–909.

Raghunath, M., Superti-Furga, A., Godfrey, M., and Steinmann, B. Syvanen, A-C., Sajantila, A., and Lukka, M. (1993). Identification of(1993). Decreased extracellular deposition of fibrillin and decorin individuals by analysis of biallelic DNA markers, using PCR andin neonatal Marfan syndrome fibroblasts. Hum. Genet. 90: 511– solid-phase minisequencing. Am. J. Hum. Genet. 52: 46–59.515. Tikkanen, R., Enomaa, N., Riikonen, A., Ikonen, E., and Peltonen,

Sakai, L. Y., Keene, D. R., and Engvall, E. (1986). Fibrillin, a new L. (1995). Intracellular sorting of aspartylglucosaminidase: The350-kD glycoprotein is a component of extracellular microfibrils. role of N-linked oligosaccharides and evidence of Man-6-P-Inde-J. Cell Biol. 103: 2499–2509. pendent lysosomal targeting. DNA Cell Biol. 14: 305–312.

Sheenan, J. K., Ratclife, A., Oates, K., and Hardingham, T. E. (1987). Vandenplas, S., Wiid, I., Grobler-Rabie, A., Brebner, K., Ricketts,The detection of substructures within proteoglycan molecules. Bio- M., Wallis, G., Bester, A., Boyd, C., and Mathew, C. (1984). Blotchem. J. 247: 267–276. hybridisation analysis of genomic DNA. J. Med. Genet. 21: 164–

171.Stacey, A., and Schnieke, A. (1990). SVpoly: A versatile mammalianexpression vector. Nucleic Acids Res. 18: 2829. Wang, M., Price, C. E., Han, J., Cisler, J., Imaizumi, K., Van Thienen,

M. N., DePaepe, A., and Godfrey, M. (1995). Recurrent mis-splicingSullivan, L. J., Makris, G. S., Dickinson, P., Mulhall, L. E., Forrest,of fibrillin exon 32 in two patients with neonatal Marfan syndrome.S., Cotton, R. G., and Loughnan, M. S. (1993). A new codon 15Hum. Mol. Genet. 4: 607 –613.rhodopsin gene mutation in autosomal dominant retinitis pig-

mentosa is associated with sectorial disease. Arch. Ophth. 111: Yamazumi, K., Shimura, K., Terukina, S., Takahashi, N., and Mat-1512–1517. suda, M. (1989). A g methionine-310 to threonine substitution and

consequent N-glycosylation atg asparagine-308 identified in a con-Superti-Furga, A., Raghunath, M., and Willems, P. J. (1992). Defi-ciencies of fibrillin and decorin in fibroblast cultures of a patient genital dysfibrinogenemia associated with posttraumatic bleeding,

fibrinogen Asahi. J. Clin. Invest. 83: 1590–1597.with neonatal Marfan syndrome. J. Med. Genet. 29: 875–878.


a point mutation creating an extran-glycosylation site in fibrillin-1 results in neonatal marfan...

Documents