Clin Genet 2012: 81: 224–233Printed in Singapore. All rights reserved

© 2011 John Wiley & Sons A/SCLINICAL GENETICS

doi: 10.1111/j.1399-0004.2011.01689.x

Original Article

Fabry disease: polymorphic haplotypesand a novel missense mutationin the GLA gene

Ferri L, Guido C, la Marca G, Malvagia S, Cavicchi C, Fiumara A,Barone R, Parini R, Antuzzi D, Feliciani C, Zampetti A, Manna R, GiglioS, Della Valle CM, Wu X, Valenzano KJ, Benjamin ER, Donati MA,Guerrini R, Genuardi M, Morrone A. Fabry disease: polymorphichaplotypes and a novel missense mutation in the GLA gene.Clin Genet 2012: 81: 224–233. © John Wiley & Sons A/S, 2011

Fabry disease (FD) is an X-linked lysosomal storage disorder with aheterogeneous spectrum of clinical manifestations that are caused by thedeficiency of α-galactosidase A (α-Gal-A) activity. Although useful fordiagnosis in males, enzyme activity is not a reliable biochemical marker inheterozygous females due to random X-chromosome inactivation, thusrendering DNA sequencing of the α-Gal-A gene, alpha-galactosidase gene(GLA), the most reliable test for the confirmation of diagnosis in females.The spectrum of GLA mutations is highly heterogeneous. Manypolymorphic GLA variants have been described, but it is unclear ifhaplotypes formed by combinations of such variants correlate with FD,thus complicating molecular diagnosis in females with normal α-Gal-Aactivity. We tested 67 female probands with clinical manifestations thatmay be associated with FD and 110 control males with normal α-Gal-Aactivity. Five different combinations of GLA polymorphic variants wereidentified in 14 of the 67 females, whereas clearcut pathogeneticalterations, p.Met51Ile and p.Met290Leu, were identified in two cases. Thelatter has not been reported so far, and both mutant forms were found tobe responsive to the pharmacological chaperone deoxygalactonojirimycin(DGJ; migalastat hydrochloride). Analysis of the male control population,as well as male relatives of a suspected FD female proband, permitted theidentification of seven different GLA gene haplotypes in strong linkagedisequilibrium. The identification of haplotypes in control males providesevidence against their involvement in the development of FD phenotypicmanifestations.

Conflict of interest

Cecilia M. Della Valle, Xiaoyang Wu, Elfrida R. Benjamin, andKenneth J. Valenzano are employees of Amicus Therapeutics andAmicus Therapeutics stockholders. All other authors have nothingto declare.

L Ferria,b, C Guidoa,b,G la Marcab, S Malvagiab,C Cavicchib, A Fiumarac,R Baronec, R Parinid,D Antuzzie, C Felicianie,A Zampettie, R Mannae,S Gigliof, CM Della Valleg, XWug, KJ Valenzanog, ERBenjaming, MA Donatib, RGuerrinib, M Genuardif and AMorronea,b

aDepartment of Sciences for Woman andChild’s Health, University of Florence,Florence, Italy, bMetabolic and MuscularUnit, Clinic of Paediatric Neurology,Meyer Childrens’ Hospital, Florence, Italy,cDepartment of Paediatrics, University ofCatania, Catania, Italy, dMetabolic Unit,San Gerardo Hospital, Monza, Italy,eDepartment of Dermatology, PoliclinicoGemelli Hospital, Catholic University,Rome, Italy, fDepartment of ClinicalPathophysiology, Medical Genetics Unit,University of Florence, Florence, Italy, andgAmicus Therapeutics, Cranbury, NJ,USA

Key words: Fabry disease – GLA genevariants – haplotypes – polymorphism

Corresponding author: Amelia Morrone,Department of Sciences for Womanand Child’s Health, Universityof Florence, Viale Pieraccini n. 24,50139 Florence, Italy.Tel.: +39 0555662543;fax: +39 0555662849;e-mail: a.morrone@meyer.it

Received 9 February 2011, revised andaccepted for publication 20 April 2011

224

Fabry disease and polymorphic haplotypes

The human lysosomal enzyme α-galactosidase A(α-Gal-A; EC3.2.1.22) hydrolyzes α-galactosidiclinkages of glycosphingolipids, glycoproteins andpolysaccharides (1). It is encoded by the alpha-galactosidase gene (GLA) gene (OMIM300644,RefSeq X14448), located on the X-chromosome(Xq22.1) (2). Total or partial deficiency of α-Gal-A activity is responsible for Fabry disease(FD; OMIM301500), a lysosomal storage disor-der (LSD) characterized by the accumulation ofglycosphingolipids within lysosomes.

Phenotypic expression of FD is highly vari-able (3–5), ranging from a less severe conditionwith only cardiac and/or renal abnormalities, tothe classic phenotype with chronic pain, vasculardegeneration, angiokeratoma, cardiac abnormali-ties, kidney manifestations leading to renal failure,and other symptoms (1). Deficient α-Gal-A activ-ity in leukocytes or fibroblasts is commonly usedto diagnose males. While heterozygous femalesmay also develop mild to severe clinical mani-festations (1), enzymatic assays are unreliable infemales because of variable X-chromosome inacti-vation. Hence, it is crucial to combine biochemicalassays with molecular analyses of GLA to identifyfemales affected with FD.

To date, more than 600 disease-causing GLAmutations (1, 6) and several benign polymor-phisms have been reported. Davies et al. (7) identi-fied three single nucleotide polymorphisms (SNPs)in the 5′ UTR, including two G>A transi-tions at nucleotides −30 (g.1150G>A) and −12(g.1168G>A) and one C>T transition at −10(g.1170C>T), as well as a 7-bp polymorphic dele-tion in intron 2, located between nucleotide posi-tions g.7184 and g.7193. The A1150 allele isrelatively rare (0.5% in the normal male pop-ulation) and was correlated with high plasmaenzyme activity, possibly because of modificationsof the binding site for transcriptional inhibitors (8).The T1170 allele was correlated with a slightreduction of α-Gal-A activity, though still in thenormal range, by both family and population stud-ies (9, 10). On the basis of these observations,Oliveira et al. (10) hypothesized that the T1170allele might increase risk of ischemic small-vesselcerebrovascular disease, although this hypothesisremains unproven.

Additional intronic benign polymorphisms havebeen identified, including g.10115A>G (IVS4-16A>G), g.10956C>T (IVS6-22C>T), and g.9401A>G (IVS4+987A>G) (11, 12). Furthermore,two deep intronic mutations, g.9331G>A (IVS4+919G>A) and g.9273C>T (IVS4+861C>T),have been associated with increased amountsof an alternative GLA transcript that is normally

weakly expressed (12, 13). Finally, the only knownnon-pathogenic amino acid substitution in thecoding region, c.937G>T (p.Asp313Tyr), has beenassociated with a pseudodeficiency (14, 15).

Recently, four females carrying three poly-morphic variants (g.1170C>T, g.10115A>G, andg.10956C>T), presumably combined in a singlehaplotype, have been identified in patients withsmall-fiber neuropathy (16). Although these sub-jects had abnormal levels of globotriaosylceramide(Gb3) and globotriaosylsphingosine (lyso-Gb3),they did not show other significant manifestationsof FD.

These findings suggest that specific GLAhaplotypes, containing combinations of multiplepolymorphic variants, may be associated withmanifestations of FD. Hence, in order to investi-gate the clinical relevance of GLA haplotypes, wedetermined α-Gal-A activity and GLA sequencein samples from 67 females that were referredbecause of suspicion of FD, and from 110 healthycontrol male subjects. The distribution of poly-morphic alleles and haplotypes was determined inthese groups and related to enzyme activity, as wellas to clinical manifestations in the female series.

Materials and methods

Patients and controls

Anonymous control samples from the Italian pop-ulation of 110 blood donor healthy males and acohort of 67 females with phenotypic manifesta-tions similar to FD (17) were examined. In addi-tion, we evaluated several relatives of one femaleproband. Informed consent was obtained from allpatients included in the study in accordance withlocal ethical committee recommendations.

α-Gal-A biochemical enzyme assay

α-Gal-A activity on dried blood spots (DBS) wasmeasured in duplicate on an API 2000 TandemMass Spectrometer (Liquid chromatography-massspectrometry (LC-MS/MS); AB Sciex, FosterCity, CA) as described (18). DBS were preparedusing filter paper (Schleicher & Schuell; RivieraBeach, FL) as reported (18). False-positive resultsdue to sample degradation were excluded bymeasuring the activities of the control lysosomalenzymes, α-glucosidase and β-glucosidase, on eachDBS sample (18). For all patients, α-Gal-A andcontrol lysosomal enzymes were also assayed oncell lysates using the 4-methylumbelliferyl-α-d-galactopyranoside (Sigma-Aldrich, St. Louis, MO)fluorogenic method as previously reported (19).

225

Ferri et al.

Our normal α-Gal-A enzyme activity ranges are asfollows: 5.7–49.7 μmol/l/h by LC-MS/MS; 20–65nmol/mg-protein/h by the fluorogenic method.

Analysis of genomic DNA

Genomic DNA was isolated from whole bloodusing the EZ1 DNA Blood 350-μl Kit and theEZ1 workstation (QIAGEN, Hilden, Germany) fol-lowing the manufacturer’s instructions. The entirecoding region and intron–exon boundaries of GLAfrom all patients and controls enrolled in thestudy were amplified by polymerase chain reaction(PCR) using reported oligonucleotides (12, 20).

After primary denaturation for 2 min at 95◦C,amplification was carried out for 30 cycles of 15 sat 95◦C, 30 s at 56◦C and 30 s at 72◦C, with afinal extension of 7 min at 72◦C. PCR productswere assessed by 2% agarose gel electrophoresisand purified by mixing 5 μl of PCR product with2 μl of Exo-SAP-IT (USB Corporation, Cleveland,OH) and incubating for 30 min at 37◦C, followedby additional 15 min at 80◦C.

Sequencing PCR reactions were performedusing the BigDye Terminator v1.1 CycleSequencing Kit (Applied Biosystems, Carlsbad,CA); purification was conducted using SephadexG-50 Fine (GE Healthcare, Little Chalfont, UK).Capillary electrophoresis was performed usingABI PRISM 3130 Genetic Analyser (AppliedBiosystems) as recommended by the manufacturer.

All nucleotide positions in this work are namedon the basis of the human GLA nucleotidesequence X14448 (21), using standard nomencla-ture (http://www.hgvs.org/mutnomen).

Multiplex ligation-dependent probe amplificationanalysis

Multiplex ligation-dependent proble amplification(MLPA) analysis was performed on DNA samplesfrom suspected FD female patients in order to iden-tify gross rearrangements of the GLA gene. MLPAwas conducted using the SALSA MLPA P159GLA kit following the manufacturer’s instructions(MRC Holland, Amsterdam, The Netherlands).

In silico analysis of missense mutations

α-Gal-A protein sequences from 16 species (Maca-ca mulatta XP_001093625.1, Oryctolagus cunicu-lus XP_002720395.1, Sus scrofa NP_001171396.1, Mus musculus NP_038491.2, Taeniopygia gut-tata XP_002187248.1, Gallus gallus XP_420183.2, Takifugu rubripes CAC44626.1, Danio rerioNP_001006103.1, Ixodes scapularis XP_0023995

73.1, Culex quinquefasciatus XP_001843334.1,Tetrahymena thermophila XP_001014570.1, As-pergillus fumigatus XP_753271.1, Escherichia coliACI73670.1, Toxoplasma gondii XP_002369542.1, Lactobacillus salivarius YP_535901.1, andYersinia pestis NP_992831.1) were aligned withhuman α-Gal-A sequence (NP_000160.1) usingthe ClustalW program (22) and multialignmentswere checked to evaluate evolutionary conserva-tion of amino acid residues involved in the mis-sense substitutions identified in patient samples.

Linkage disequilibrium analysis

The value D′ of linkage disequilibrium (LD)between GLA polymorphisms was estimated ac-cording to Lewontin (23).

Construction and expression of c.153G>A(p.Met51Ile) and c.868A>C (p.Met290Leu) mutantGLA-cDNA plasmids

Mutations p.Met51Ile and p.Met290Leu were gen-erated using PCR-based site-directed mutagenesisof GLA-cDNA, inserted into the expression vec-tor pcDNA6 (Invitrogen, Carlsbad, CA). Forwardprimers: 5′-GCGCTTCATATGCAACC-3′ (p.Met51Ile) and 5′-CCTCTGGGCTATCCTGGCTGCTCCTTTA-3′ (p.Met290Leu) and reverse primers: 5′-GGTTGCATATGAAGCGC-3′ (p.Met51Ile) and5′-TAAAGGAGCAGCCAGGATAGCCCAGAGG-3′ (p.Met290Leu) were used for the mutage-nesis reactions according to the standard proto-cols. Mutations were confirmed by sequencing thefull-length GLA-cDNA. Plasmid preparations weremade using the PureYield™ Plasmid Maxiprep Kit(Promega, Madison, WI).

Effect of DGJ on α-Gal-A activity

HEK-293 cells (GripTite-293 MSR-cells; Invit-rogen) were seeded in 96-well plates (CorningLife Sciences, Lowell, MA) at a density of 7500cells/well and incubated at 37◦C, 5% CO2 for24 h. Cells were transfected with empty vectoror mutant α-Gal-A-plasmids using reagent FugeneHD (Roche, Basel, Switzerland) according to man-ufacturer’s instructions. After transfection, cellswere incubated at 37◦C, 5% CO2 for 1 h beforeadding 1-deoxygalactonojirimycin (DGJ; migalas-tat; Sigma-Aldrich) at concentrations ranging from50 nM to 1 mM. Cells were then incubated for4–5 days before lysis and assay. α-Gal-A activitywas assayed as described previously (24).

T-cell-based assays were performed by incu-bating patient-derived lymphocytes with 0, 20,

226

Fabry disease and polymorphic haplotypes

or 100 μM DGJ for 5 days at 37◦C in 5%CO2. After incubation, cells were harvested andwashed four times with 1 ml 0.9% NaCl solution.Lymphocyte pellets were sonicated 10 s andassayed in triplicate for α-Gal-A activity and bywestern blot.

Western blot analysis

Thirty micrograms of lysates were loaded per well.Blots were prepared from 12.5% polyacrylamidegels and probed as previously described (25).After electrophoresis, proteins were transferredonto nitrocellulose (Bio-Rad, Hercules, CA) andfilters were incubated with anti-α-Gal-A antibod-ies (diluted 1:200) kindly provided by Shire Cor-poration (Shire, Italy). Blots were then incubatedwith secondary anti-rabbit IgG (whole moleculediluted 1:5000) conjugated to alkaline phosphatase(Sigma, Milano, Italy) and signals were revealedusing the AP Conjugate Substrate kit (Bio-Rad).

Results

Identification of GLA variants in female probands withFD manifestations

GLA variants were identified in 14 of the 67females referred for genetic testing due to clinicalmanifestations resembling FD (Table 1). α-Gal-Aactivity was normal in 66 subjects and was reducedin 1 subject (Pt3 in Table 1).

In addition, three missense changes were ob-served (each with a frequency of 1/67; 1.5%): thepseudodeficiency variant p.Asp313Tyr (14, 15),c.153G>A transition (p.Met51Ile) (26) (Pt5) anda novel c.868A>C transversion (p.Met290Leu).No GLA gross rearrangements were identified byMLPA.

Five different combinations of GLA polymor-phisms in non-coding regions were identified inthese females (Table 2). The most frequent wasg.1170C>T + g.7188-7192del5 + g.10115A>G +g.10956C>T, detected in 9 of the 67 (13.4%)probands.

Enzyme activity and GLA variants in the Italian malecontrol population

α-Gal-A activity, measured by LC-MS/MS, wasnormal in all 110 male controls examined.Activities of the control enzymes β-glucosidase(n.v. 1.7–38.3 μmol/l/h) and α-glucosidase (n.v.1.9–21.9 μmol/l/h) were also within normalranges. Several polymorphisms were observedin these control subjects by GLA sequencing(Table 3), including the previously reported g.1168

G>A, g.1170C>T, g.10115A>G and g.10956C>T (7, 11), as well as the deletion g.7188-7192del5 within intron 2 (rs5903184). This poly-morphism is very similar to the previouslydescribed 7-bp deletion (7), but in all our controlsit was identified as a 5-bp deletion. In addi-tion, a novel polymorphism, g.9278-9279del2(IVS4+866-867del2), was found in intron 4 of onemale control.

GLA haplotype combinations

GLA non-coding polymorphisms were combinedin five different haplotypes in the control popula-tion analyzed (Table 4). These were labeled withabbreviation GH (GLA Haplotype) followed by aprogressive number. Haplotypes GH1–GH5 corre-sponded to the combinations of polymorphic vari-ants detected in the females (Table 2).

In order to determine the segregation of GLApolymorphisms we also examined several mem-bers of a Sicilian family (Fig. 1) ascertainedthrough a female proband (III.4 in Fig. 1; Pt9 inTable 1). Two haplotype combinations, GH4 andGH6, were clearly identified in this proband. TheGH6 haplotype, which was not observed amongmale controls, was shared by III.4 and her twobrothers (III.1 and III.2), both of whom had nor-mal α-Gal-A activity, as well as her mother (II.2)and maternal grandfather (I.1). An additional hap-lotype not found in the control population, GH7,was identified in two maternal uncles (II.5 andII.6), both of whom had normal α-Gal-A activity,as well as in maternal grandmother (I.2).

LD was estimated for the most common combi-nations of two variants: g.1168G>A-g.10956C>T(present in haplotypes GH1 and GH3), g.7188-7192del5-g.10956C>T (present in haplotypesGH2, GH4 and GH5), and g.1170C>T-g.10956C>T (present in haplotypes GH4 and GH5). Allcombinations were found to be in complete LD inthe male population (D′ = 1).

In silico analysis of missense mutations

ClustalW analysis showed that p.Met290 inthe human α-Gal-A protein is phylogeneticallyconserved in mammals and other vertebrates,whereas it is different in insects and bacteria (datanot shown). In contrast, p.Met51 on the humanprotein is not phylogenetically conserved amongmammals (data not shown).

In vitro expression of mutant α-Gal-A and response tothe pharmacological chaperone DGJ

We tested the ability of DGJ to rescue p.Met290Leu and p.Met51Ile α-Gal-A (26). In lymphocyte

227

Ferri et al.

Tabl

e1.

Clin

ical

char

acte

ristic

san

dα-G

al-A

activ

ityin

fem

ale

prob

ands

with

GLA

-var

iant

geno

type

com

bina

tions

(Y=

yes;

N=

no)

Pat

ient

12

34

56

78

910

1112

1314

Orig

inIta

lyIta

lyIta

lyV

enez

uela

Italy

Italy

Italy

Italy

Italy

Ger

man

yIta

lyIta

lyIta

lyIta

lyA

geof

susp

icio

nof

FD18

2434

1018

——

713

3964

4032

10

Ang

ioke

rato

ma

NN

NN

NN

NN

NN

NN

NN

Mus

culo

skel

etal

and

bone

man

ifest

atio

ns

Y(o

steo

necr

osis

offe

mor

alhe

ad)

NN

NN

NN

NN

NN

NN

N

Cor

nea

vert

icilla

taN

YN

NN

NN

NN

NN

NN

ND

iaph

ores

isN

NN

NN

NN

NN

NN

NN

NA

bdom

inal

pain

NN

NY

(with

chro

nic

cons

tipat

ion)

Y(e

piga

stric

pyro

sis,

occa

sion

alco

nstip

a-tio

n)

YN

NY

NN

NN

N

Dia

rrhe

aN

NY

NN

NN

NY

—N

NN

NTi

nnitu

san

d/or

hear

ing

loss

NN

Tinn

itus

(occ

a-si

onal

ly)

NN

NN

NN

Hea

ring

loss

NH

earin

glo

ssN

N

Acr

op-

ares

thes

ias

Y(h

ands

and

feet

,ex

acer

bate

dby

expo

sure

tow

arm

,but

not

toco

ldte

mpe

r-at

ures

)

NN

Y(h

ands

and

feet

)N

NY

(han

ds)

Y(h

ands

and

feet

)N

NN

NN

Y(le

ftha

ndan

dbo

thfe

et)

Feve

rpa

incr

ises

NN

NN

YN

NN

NN

NN

NN

Car

diov

ascu

lar

man

ifest

atio

nsN

NY

NN

NN

NN

NN

NN

N

Cer

ebro

vasc

ular

man

ifest

atio

nsN

NN

NN

Y(s

trok

e)Y

(str

oke)

NN

Y(s

trok

e)Y

(str

oke)

Y(s

trok

e)Y

(str

oke)

N

Psy

chia

tric

sym

ptom

sN

NY

NY

NN

NN

NN

NN

N

Val

vedi

seas

eN

NN

NN

NN

NN

NN

NN

NE

CG

abno

rmal

-iti

esN

NN

NN

NN

Y(fo

calr

ight

bund

lebr

anch

bloc

k)

NN

NN

NN

Hyp

erte

nsio

nN

NN

NN

NN

NN

NN

NN

NC

ardi

omyo

path

yN

NY

YN

NN

NN

NN

NN

NR

enal

man

ifes-

tatio

nsN

NY

NN

NN

NN

NN

NN

N

Oth

erN

NN

NY

(mig

rain

e)N

NN

NY

(mig

rain

e,th

rom

bosi

s,dy

spha

gia,

dysa

rthr

ia)

NN

NN

α-G

al-A

enzy

me

activ

ity(V

.N.

20–

65nm

ol/

mg/

h)

Not

done

3711

3533

2932

2427

3435

2530

32

Gen

otyp

eaG

H1

GH

1G

H2

+c.

868

A>

C(p

.Met

290L

eu)

GH

3G

H4

+c.

153

G>

A(p

.Met

51Ile

)

GH

4G

H4

GH

4G

H4

+G

H6

GH

4G

H4

GH

4c.

937

G>

Tp.

Asp

313T

yrG

H4

α-G

al-A

,α-g

alac

tosi

dase

A;E

CG

,ele

ctro

card

iogr

am;F

D,F

abry

dise

ase.

aC

ombi

natio

nsof

GLA

varia

ntal

lele

sar

ein

dica

ted

asha

plot

ype

num

ber

asde

term

ined

inth

em

ale

popu

latio

n.

228

Fabry disease and polymorphic haplotypes

Table 2. Polymorphisms and polymorphic allele combinationsin GLA non-coding regions identified in 14 females with Fabrydisease manifestations

Allele combinations

g.1168 G>A + g.10956 C>T 2g.7188-7192 del5 + g.10956 C>T 1a

g.1168 G>A + g.9278-9279 del2 + g.10956 C>T 1g.1170 C>T + g.7188-7192 del5 + g.10115 A>G +

g.10956 C>T8b

g.1170 C>T + g.7188-7192 del5 (homozygous) +g.10115 A>G (homozygous) + g.10956 C>T(homozygous)

1

c.937 G>T (p.Asp313Tyr) 1

aThis patient (Pt3 in Table 1) also carries the missense nucleotidechange c.868 A>C (p.Met290Leu).bOne of these patients (Pt 5 in Table 1) also carries the missensenucleotide change c.153 G>A (p.Met51Ile).

Table 3. GLA polymorphic variants in the Italian male controlpopulation

Polymorphism Location N (%) References

g.1168 G>A 5′ untranslatedregion

6 (5.45) (7)

g.1170 C>T 5′ untranslatedregion

4 (3.63) (7)

g.7188-7192 del5 Intron 2 9 (8.18) (7)g.9278-9279 del2 Intron 4 1 (0.9) This workg.10115 A>G Intron 4 9 (8.18) (11)g.10956 C>T Intron 6 15 (13.63) (11)

cultures established from Pt3 (heterozygous forp.Met290Leu), 20 μM DGJ, the optimal con-centration for many other mutant forms ofα-Gal-A (24, 27), increased the quantity ofdetectable enzyme by western blot, as well asthe measured enzyme activity (from approxi-mately 25% to within 88% of control; Fig. 2).The absolute mean value of enzymatic activity inthe presence of DGJ (44 nmol/mg/h) was withinnormal range (20–65 nmol/mg/h). We couldnot test a cell line from Pt5, heterozygous for

p.Met51Ile, as she refused any further investiga-tion. As FD is X-linked, female patient-derivedcell populations contain a mixture of cells thatexpress either wild-type or mutant α-Gal-A.Hence, baseline activity and the effect of DGJ onthe mutant form are difficult to accurately deter-mine. To circumvent this, our second approachdetermined p.Met51Ile and p.Met290Leu α-Gal-A responses following heterologous expressionin HEK-293 cells (Fig. 3). Incubation with DGJresulted in concentration-dependent increases ofα-Gal-A activity, attaining maximum relativeincreases of 2.6 ± 0.5-fold (n = 3) and 1.6± 0.1-fold (n = 4) above baseline, respectively.The concentration of DGJ that mediated 50%of the maximal response for p.Met51Ile andp.Met290Leu α-Gal-A was 2.3 ± 0.2 μM (n = 3)and 8.6 ± 1.3 μM (n = 4), respectively.

Discussion

More than 600 pathogenetic mutations (HumanGene Mutation Database (HGMD), http://www.hgmd.cf.ac.uk/ac/index.php) and many SNPs(http://www.ncbi.nlm.nih.gov/SNP) have been iden-tified in human GLA. SNPs in genes associatedwith Mendelian disorders that have frequency>1% are usually considered neutral. Occasion-ally, however, a common variant may have weakfunctional effects that can modulate the conse-quences of pathogenetic mutations. In the LSDGM1-gangliosidosis, the haplotype containing thep.Leu436Phe SNP in CIS with the p.Arg201Cysmutation markedly reduces β-galactosidase activ-ity, increasing phenotypic severity compared top.Arg201Cys alone (28).

Polymorphic alleles that are associated withpseudodeficiencies are relatively common in lyso-somal enzymes, such as arylsulfatase A (29),α-glucosidase (30), β-glucuronidase (31, 32), α-fucosidase (33), α-iduronidase (34), β-hexosa-minidase (35), β-galactosidase (36) and α-galacto-

Table 4. New GLA polymorphic haplotypes identified

Name Haplotype Frequency (%)

Polymorphic haplotypes in the Italian male control populationGH1 g.1168 G>A + g.10956 C>T 5 (4.54)GH2 g.7188-7192 del5 + g.10956 C>T 5 (4.54)GH3 g.1168 G>A + g.9278-9279 del2 + g.10956 C>T 1 (0.9)GH4 g.1170 C>T + g.7188-7192 del5 + g.10115 A>G + g.10956 C>T 1 (0.9)GH5 g.1170 C>T + g.7188-7192 del5 + g.10956 C>T 3 (2.72)

Polymorphic haplotypes found in male with normal α-Gal-A activity relatives of Pt 9GH6 g.7188-7192 del5 + g.10115 A>G + g.10956 C>T —GH7 g.1170 C>T + g.9151 C>T + g.10956 C>T —

α-Gal-A, α-galactosidase A.

229

Ferri et al.

I.1 I.2

III.1

II.2 II.3 II.4 II.5 II.6

III.2 III.4III.3 III.5 III.6 III.7 III.8

II.1

GH7

GH7 GH7GH4 GH6

GH6

GH6 GH6 GH4 GH4+GH6

Fig. 1. Pedigree of the Pt9 family. The arrow indicates the proband with Fabry disease manifestations (III.4). Haplotypes areindicated for individuals subjected to molecular analysis.

a

b

Fig. 2. α-galactosidase A (α-Gal-A) enzyme activity andwestern blot analyses of primary lymphocyte cultures incubatedwith deoxygalactonojirimycin (DGJ). (a) Representative α-Gal-A activity (expressed as nmol 4-MU released/mg-protein/h)in lysates from T-lymphocyte cultures incubated with theindicated concentrations of DGJ. Each value of α-Gal-Aactivity is the mean ± SD of three independent experiments.(b): Western blot analysis of α-Gal-A protein in lymphocytecultures. Blots were conducted using 30 μg total protein, theanti-α-Gal-A antibody, and an anti-β-actin antibody as a loadingcontrol.

sidase A (14, 15, 37). The difficulty to distinguishneutral polymorphisms from functionally and clin-ically relevant alleles is highlighted by a recentreport demonstrating the benign role of a common

Fig. 3. p.Met51Ile and p.Met290Leu α-galactosidase A (α-Gal-A) activity in transiently transfected HEK-293 cells andresponse to incubation with deoxygalactonojirimycin (DGJ).Representative mutant α-Gal-A activity in lysates that wereprepared from transiently transfected HEK-293 cells followingincubation with increasing concentrations of DGJ is shown. Theα-Gal-A activity was calculated as nanomoles of 4-MU releasedper milligram protein per hour (nmol/mg-protein/h) aftersubtraction of the α-Gal-A activity measured in empty vector-transfected HEK-293 cells at each DGJ concentration. Theα-Gal-A activity was normalized to the baseline of each mutantform in the absence of DGJ (defined as 1). Each data point is themean ± SEM of quadruplicate determinations. The p.Met51Ileand p.Met290Leu α-Gal-A baseline activities were 6300 ± 238and 22930 ± 3655 nmol/mg-protein/h, respectively; maximalDGJ-stimulated α-Gal-A activities were 16389 ± 2370 and37581 ± 7456 nmol/mg-protein/h, respectively (mean ± SEMfor n = 3 and n = 4 experiments, respectively). The α-Gal-A activity in wild-type α-Gal-A transfected HEK-293 cells(incubated in the absence of DGJ) was 38676 ± 2674 nmol/mg-protein/h (mean ± SEM; n = 51).

missense variant of the Arylsulfatase B (ARSB)gene, previously described as a disease-causingmutation in Maroteaux–Lamy syndrome (38). Arelevant example for α-Gal-A is p.Asp313Tyr,which causes pseudodeficiency (14, 15), but wasinitially described as a disease-causing muta-tion (39).

230

Fabry disease and polymorphic haplotypes

Absent or reduced α-Gal-A activity can be easilydetected in Fabry male patients, whereas heterozy-gous females require molecular testing because ofrandom X-chromosome inactivation. FD is sus-pected often in females, and in some who havenormal enzyme activity and lack clearcut patho-genetic alterations, different combinations of GLApolymorphic nucleotide variants can be detected.

Several studies have investigated potential func-tional effects of GLA polymorphic variants.Oliveira et al. (10) suggested that the g.1170C>TSNP may contribute to the risk of ischemicsmall-vessel cerebrovascular disease in the contextof multifactorial inheritance. Tanislav et al. (16)hypothesized that the haplotype (g.1170C>T +g.10115A>G + g.10956C>T) might alter Gb3/lyso-Gb3 levels in four females with small-fiberneuropathy. To date, these hypotheses remainunproven and the clinical relevance, if any, of GLApolymorphisms is unknown.

In the females investigated in this study withclinical suspicion of FD, the frequency of poly-morphic variants was 20.9%. Two of the 14 sub-jects carrying GLA polymorphisms also had therare coding nucleotide substitutions c.153G>Aand c.868A>C that lead to the p.Met51Ile andp.Met290Leu missense changes, respectively. Tothe best of our knowledge, the latter has not beenpreviously reported.

Although p.Met51 is not a phylogeneticallyconserved amino acid residue of α-Gal-A, it isimportant for interactions with p.Trp47 in theactive site (40). p.Met51Ile and p.Met51Lys alter-ations, which introduce a non-polar and polarresidue, respectively, were reported (26, 41). Inter-estingly, p.Met51Ile was detected, by neonatalnewborn screening, in an Italian proband whosematernal grandfather had mild hypertrophic car-diomyopathy (26). The female patient (Pt5 inTable 1) identified in this study presented withunexplained recurrent fever. Although the intactstructure of the active site is crucial for enzymeactivity, both p.Met51Lys and p.Met51Ile havebeen reported to respond to the pharmacologicalchaperone DGJ (24, 26, 27), with the responsive-ness of p.Met51Ile confirmed in the present study.

The new substitution p.Met290Leu (Pt3) in-volves a phylogenetically conserved residue inmammals and other vertebrates. Another ami-noacid change involving the same codon, p.Met290Ile, was reported in FD (6). It has been suggestedthat alterations of p.Met290 residue, located ina hydrophobic pocket, could destabilize the ter-tiary structure of α-Gal-A (6). In agreement withthis hypothesis, our results show that p.Met290Leuleads to lower α-Gal-A activity compared to

wild type when transiently overexpressed inHEK-293 cells. Moreover, our data indicate thatp.Met290Leu responds well to DGJ and can attainnear wild-type levels at DGJ concentrations thatcan be achieved clinically (42).

The other 12 females with GLA polymorphicvariants did not have additional genetic alterations.In order to establish whether GLA polymorphismsoccurred in specific haplotypes and to determinetheir frequencies in the general population, weinvestigated 110 control male samples with α-Gal-A activity in normal range. We identified fivedifferent polymorphic alleles of GLA that sharedstrong LD. Four of the five haplotypes identified inthe male control group corresponded to genotypecombinations observed in the female series, indi-cating that these were probably present as haplo-type combinations in female patients. Variant hap-lotypes did not influence α-Gal-A activity in themale control series, as determined by the compar-ison of genotype/enzymatic data (data not shown).

Two other GLA haplotypes were found bythe analyses of a pedigree ascertained throughfemale proband Pt9 (Fig. 1). This proband washeterozygous for g.1170C>T and homozygous forg.7192-7198del5, g.10115A>G and g.10956C>T.Molecular analyses of her parents showed that hergenotype was formed by the combination of twohaplotypes, GH4 and GH6. GH6, which had notbeen observed among controls, was also present inher two brothers and maternal grandfather who hadnormal α-Gal-A activity; these findings indicatethat GH6 has no major effect on enzyme function.The same applies to the other novel haplotypeidentified in this family, GH7, that was observedin two maternal uncles (Fig. 1), both of whom hadnormal α-Gal-A activity.

In conclusion, this work describes for the firsttime the distribution of polymorphic GLA haplo-types. The observation that these do not correlatewith significant variations of α-Gal-A activity indi-cates that they are not likely to have major pheno-typic effects, although we cannot completely ruleout weak effects that could be detected on largerseries. However, at present, the detection of poly-morphic haplotype combinations, including lessfrequent GLA polymorphic alleles, in the absenceof other GLA sequence variations in patients withsuspected FD should be considered as a coinciden-tal finding, not relevant for FD diagnosis.

Acknowledgements

This work was partially supported by grants from the GenzymeCorporation, Ateneo (MIUR ex 60%) and Shire Italia. Anti-α-Gal-A antibody was kindly provided by Shire Corporation. DGJ

231

Ferri et al.

was purchased from Sigma–Aldrich (St. Louis, MO) and waskindly provided by Amicus Therapeutics (Cranbury, NJ). AMMeC(Associazione Malattie Metaboliche Congenite, Italia) and AIMPS(Associazione Italiana Mucopolisaccaridosi e malattie affini, Italia)are also kindly acknowledged for their support.

References

1. Desnick RJ, Ioannou YA, Eng CM. α-galactosidase A defi-ciency: Fabry disease. In: Scriver CR, Beaudet AL, Sly WS,Valle D, eds. The metabolic and molecular bases of inher-ited disease, 8th edn. New York, NY: McGraw-Hill Inc, 2001:3733–3774.

2. Vetrie D, Kendall E, Coffey A et al. A 6.5-Mb yeast artificialchromosome contig incorporating 33 DNA markers on thehuman X chromosome at Xq22. Genomics 1994: 19 (1):42–47.

3. Elleder M, Bradova V, Smíd F et al. Cardiocyte storageand hypertrophy as a sole manifestation of Fabry’s disease.Report on a case simulating hypertrophic non-obstructivecardiomyopathy. Virchows Arch A Pathol Anat Histopathol1990: 417 (5): 449–455.

4. Frustaci A, Chimenti C, Ricci R et al. Improvement incardiac function in the cardiac variant of Fabry’s diseasewith galactose-infusion therapy. N Engl J Med 2001: 345 (1):25–32.

5. Nakao S, Kodama C, Takenaka T et al. Fabry disease: detec-tion of undiagnosed hemodialysis patients and identificationof a “renal variant” phenotype. Kidney Int 2003: 64 (3):801–807.

6. Shabbeer J, Yasuda M, Benson SD, Desnick RJ. Fabry dis-ease: identification of 50 novel α-galactosidase mutations caus-ing the classic phenotype and three-dimensional structuralanalysis of 29 missense mutations. Hum Genomics 2006: 2:297–309.

7. Davies JP, Winchester BG, Malcom S. Sequence variations inthe first exon of α-galactosidase A. J Med Genet 1993: 30:658–663.

8. Fitzmaurice TF, Desnick RJ, Bishop DF. Human alpha-galactosidase A: high plasma activity expressed by the-30G–>A allele. J Inherit Metab Dis 1997: 20 (5): 643–657.

9. Oliveira JP, Ferreira S, Barcelo J et al. Effect of single-nucleotide polymorphisms of the 5’ untranslated region ofthe human alpha-galactosidase gene on enzyme activity, andtheir frequencies in Portuguese caucasians. J Inherit Metab Dis2008. Available at http://dx.doi.org/10.1007/s10545-008-0818-9. Accessed on November 30, 2008.

10. Oliveira JP, Ferreira S, Reguenga C, Carvalho F, Mansson JE.The g.1170C>T polymorphism of the 5’ untranslated region ofthe human alpha-galactosidase gene is associated withdecreased enzyme expression-Evidence from a family study.J Inherit Metab Dis 2008. Available at http://dx.doi.org/10.1007/s10545-008-0972-0. Accessed on November 30, 2008.

11. Davies JP, Winchester BG, Malcolm S. Mutation analysis inpatients with the typical form of Anderson-Fabry disease. HumMol Genet 1993: 2 (7): 1051–1053.

12. Ishii S, Nakao S, Minamikawa-Tachino R, Desnick RJ, FanJQ. Alternative splicing in the alpha-galactosidase gene:increased exon inclusion results in the Fabry cardiac pheno-type. Am J Hum Genet 2002: 70: 994–1002.

13. Filoni C, Caciotti A, Carraresi L et al. Unbalanced GLAmRNAs ratio quantified by real-time PCR in Fabry patients’fibroblasts results in Fabry disease. Eur J Hum Genet 2008:16 (11): 1311–1317.

14. Froissart R, Guffon N, Vanier MT, Desnick RJ, Maire I. Fabrydisease: D313Y is an alpha-galactosidase A sequence variant

that causes pseudodeficient activity in plasma. Mol GenetMetab 2003: 80 (3): 307–314.

15. Yasuda M, Shabbeer J, Benson SD, Maire I, BurnettRM, Desnick RJ. Fabry disease: characterization of alpha-galactosidase A double mutations and the D313Y plasmaenzyme pseudodeficiency allele. Hum Mutat 2003: 22 (6):486–492.

16. Tanislav C, Kaps M, Rolfs A et al. Frequency of Fabry diseasein patients with small-fibre neuropathy of unknown aetiology:a pilot study. Eur J Neurol 2011: 18 (4): 31–36.

17. Beck M. The Mainz Severity Score Index (MSSI): develop-ment and validation of a system for scoring the signs andsymptoms of Fabry disease. Acta Paediatr Suppl 2006: 95(451): 43–46.

18. la Marca G, Casetta B, Malvagia S, Guerrini R, Zammarchi E.New strategy for the screening of lysosomal storage disor-ders: the use of the online trapping-and-cleanup liquid chro-matography/mass spectrometry. Anal Chem 2009: 81 (15):6113–6121.

19. Filoni C, Caciotti A, Carraresi L et al. Functional studies ofnew GLA gene mutations leading to conformational Fabrydisease. Biochim Biophys Acta 2010: 1802 (2): 247–252.

20. Morrone A, Cavicchi C, Bardelli T et al. Fabry disease:molecular studies in Italian patients and X inactivation analysisin manifesting carriers. J Med Genet 2003: 40 (8): e103.

21. Kornreich R, Desnick RJ, Bishop DF. Nucleotide sequence ofthe human alpha-galactosidase A gene. Nucleic Acids Res1989: 17 (8): 3301–3302.

22. Thompson JD, Higgins DG, Gibson TJ. CLUSTAL W:improving the sensitivity of progressive multiple sequencealignment through sequence weighting, position-specific gappenalties and weight matrix choice. Nucleic Acids Res 1994:22: 4673–4680.

23. Lewontin RC. The interaction of selection and linkage.I. General considerations; heterotic models. Genetics 1964: 49:49–67.

24. Benjamin ER, Flanagan JJ, Schilling A et al. The pharmaco-logical chaperone 1-deoxygalactonojirimycin increases alpha-galactosidase A levels in Fabry patient cell lines. J InheritMetab Dis 2009: 32 (3): 424–440.

25. Van Dongen JM, Willemsen R, Ginns EI et al. The subcel-lular localization of soluble and membrane-bound lysosomalenzymes in I-cell fibroblasts: a comparative immunocytochem-ical study. Eur J Cell Biol 1985: 39: 179–189.

26. Spada M, Pagliardini S, Yasuda M et al. High incidence oflater-onset Fabry disease revealed by newborn screening. AmJ Hum Genet 2006: 79 (1): 31–40.

27. Shin SH, Kluepfel-Stahl S, Cooney AM et al. Prediction ofresponse of mutated alpha-galactosidase A to a pharmaco-logical chaperone. Pharmacogenet Genomics 2008: 18 (9):773–780.

28. Caciotti A, Bardelli T, Cunningham J, d’Azzo A, ZammarchiE, Morrone A. Modulating action of the new polymorphismL436F detected in the GLB1 gene of a type-II GM1gangliosidosis patient. Hum Genet 2003: 113 (1): 44–50.

29. Gieselmann V, Polten A, Kreysing J, von Figura K. Arylsulfa-tase A pseudodeficiency: loss of a polyadenylation signal andN-glycosylation site. Proc Natl Acad Sci U S A 1989: 86:9436–9440.

30. Nishimoto J, Inui K, Okada S et al. A family with pseu-dodeficiency of acid a-glucosidase. Clin Genet 1988: 33:254–261.

31. Chabas A, Giros ML, Guardiola A. Low β-glucuronidaseactivity in a healthy member of a family with mucopolysac-charidosis VII. J Inherit Metab Dis 1991: 14: 908–914.

232

Fabry disease and polymorphic haplotypes

32. Vervoort R, Islam MR, Sly W et al. A pseudodeficiency allele(D152N) of the human beta-glucuronidase gene. Am J HumGenet 1995: 57: 798–804.

33. Wauters JG, Stuer KL, Elsen AV, Willems PJ. α-L-fucosidasein human fibroblasts. I. The enzyme activity polymorphism.Biochem Genet 1992: 30: 131–141.

34. Aronovich EL, Pan D, Whitley CB. Molecular genetic defectunderlying alpha-L-iduronidase pseudodeficiency. Am J HumGenet 1996: 58: 75–85.

35. Cao Z, Petroulakis E, Salo T, Triggs-Raine B. Benign HEXAmutations, C739T(R247W) and C745T(R249W), cause beta-hexosaminidase A pseudodeficiency by reducing the alpha-subunit protein levels. J Biol Chem 1997: 272: 14975–14982.

36. Gort L, Santamaria R, Grinberg D, Vilageliu L, Chabas A.Identification of a novel pseudodeficiency allele in the GLB1gene in a carrier of GM1 gangliosidosis. Clin Genet 2007: 72:109–111.

37. Hoffmann B, Georg Koch H, Schweitzer-Krantz S, Wendel U,Mayatepek E. Deficient alpha-galactosidase A activity inplasma but no Fabry disease–a pitfall in diagnosis. Clin ChemLab Med 2005: 43: 1276–1277.

38. Zanetti A, Ferraresi E, Picci L et al. Segregation analysis in afamily at risk for the Maroteaux-Lamy syndrome conclusivelyreveals c.1151G>A (p.S384N) as to be a polymorphism. EurJ Hum Genet 2009: 17 (9): 1160–1164.

39. Eng CM, Resnick-Silverman LA, Niehaus DJ, Astrin KH,Desnick RJ. Nature and frequency of mutations in the alpha-galactosidase A gene that cause Fabry disease. Am J HumGenet 1993: 53 (6): 1186–1197.

40. Garman SC, Garboczi DN. The molecular defect leading toFabry disease: structure of human alpha-galactosidase. J MolBiol 2004: 337 (2): 319–335.

41. Ashley GA, Shabbeer J, Yasuda M, Eng CM, Desnick RJ.Fabry disease: twenty novel alpha-galactosidase A mutationscausing the classical phenotype. J Hum Genet 2001: 46 (4):192–196.

42. Schiffmann R, Germain DP, Castelli J, Shenker A, LockhartDJ. Phase 2 clinical trials of the pharmacological chaperoneAT1001 for the treatment of Fabry disease. Presented at theAmerican Society of Human Genetics Conference, Philadel-phia, PA, November 11–15, 2008. Available at http://www.ashg.org/2008meeting. Accessed on November 30, 2008.

233