www.elsevier.com/locate/ymgme

Molecular Genetics and Metabolism 86 (2005) 220–232

Niemann–Pick C disease: Use of denaturing high performanceliquid chromatography for the detection of NPC1 and NPC2

genetic variations and impact on managementof patients and families q

Gilles Millat *, Nathalie Baılo, Sabine Molinero, Celine Rodriguez,Karim Chikh, Marie T. Vanier

Inserm and Fondation Gillet-Merieux, Lyon-Sud Hospital, 69310 Pierre-Benite, France

Received 27 June 2005; received in revised form 13 July 2005; accepted 14 July 2005Available online 26 August 2005

Abstract

Niemann–Pick disease type C (NPC), a neurovisceral disorder characterized by accumulation of unesterified cholesterol andglycolipids in the lysosomal/late endosomal system, is due to mutations on either the NPC1 or the NPC2 genes. While the corre-sponding proteins appear essential for proper cellular cholesterol trafficking, their precise function and relationship are still unclear.Mutational analysis of patients, useful for the study of structure/function relationships, is especially valuable for proper manage-ment of affected families. Correlations have been found between genotypes and the severity of the neurological outcome of thepatients, and molecular genetics constitutes the optimal approach for prenatal diagnosis. However, mutation detection in NPC dis-ease is a challenge. The NPC1 gene, affected in >95% of the families, is large in size (�50 kb), and the already known disease-caus-ing mutations and numerous polymorphisms are scattered over 25 exons. Furthermore, detection of NPC2 patients by complexgenetic complementation tests is unpractical. In the present study, we describe a rapid and reliable strategy for detecting NPCgenetic variations using DHPLC analysis. Conditions of analysis were optimized for all the NPC1 and NPC2 30 exons and val-idated using 38 previously genotyped patients. These conditions were then applied to screen a panel of 35 genetically uncharacter-ized, unrelated NPC patients. Pathogenic mutations were identified in 68/70 alleles. Among the mutations identified, 29 were novel,including two of the NPC2 gene. We conclude that DHPLC is a rapid, low-cost, highly accurate, and efficient technique for thedetection of NPC genetic variants.� 2005 Elsevier Inc. All rights reserved.

Keywords: Niemann–Pick disease; NPC2; NPC1; Denaturing high-performance liquid chromatography; DHPLC; Mutation detection

Introduction

Niemann–Pick type C disease (NPC, OMIM 257220)is a rare autosomal recessive lipid disorder usually char-

1096-7192/$ - see front matter � 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.ymgme.2005.07.007

q Databases: NPC1-OMIM, #607623; GenBank: NM_000271,NPC2-OMIM, #601015; GenBank: NM_640032.3.* Corresponding author. Fax: +33 4 78505494.E-mail address: millat@lyon.inserm.fr (G. Millat).

acterized by hepatosplenomegaly and progressive neuro-logical deterioration with varying age at onset andvarying later course. In cultured cells, the most promi-nent biochemical feature is a lysosomal/late endosomalaccumulation of endocytosed unesterified cholesteroland delayed induction of cholesterol homeostatic reac-tions [1–4]. Marked variations have been observed inthe severity of the cellular cholesterol lesion: typical se-vere alterations described as the ‘‘classic’’ biochemical

G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232 221

phenotype have been observed in most patients, andmild alterations reported as the ‘‘variant’’ phenotypein about 20% of them [5]. A considerable progress hasbeen achieved during the last decade, with the recogni-tion that mutations in two different genes, eitherNPC1 or NPC2, can cause the disease [6,7], the identifi-cation of the genes [8,9], and increasing knowledge onthe gene products. Nevertheless, the exact cellular func-tion of NPC1 and NPC2 proteins and their role in theintracellular trafficking of lipids remain unclear.

The NPC1 gene, mutated in 95% of NPC families,encodes an integral membrane protein predominantlylocated in late endosomes [10–12]. To date, nearly200 different disease-causing NPC1 mutations havebeen reported, with a large majority (ca. 70%) of mis-sense mutations [8,13–30]. Only three frequent NPC1mutations have been described. Mutation p.I1061T,associated with a ‘‘classic’’ biochemical form, accountsfor approximately 20% of the alleles in the UnitedKingdom and France and 15% in the USA [15]. Thetwo other most recurrent mutations are associated witha ‘‘variant’’ biochemical form: p.P1007A, the secondmost frequent mutation in Europe, and p.G992W typ-ical of Nova-Scotian patients but rare in other popula-tions [4]. Genotype–phenotype correlations in NPC1patients have previously suggested that three NPC1 do-mains are functionally critical: the sterol-sensing do-main (SSD), the large luminal cysteine-rich loop, andthe luminal ‘‘NPC1 domain.’’ More than 50 exonicand intronic SNPs have also been described [16–18,21,23,25,26,30]. The most prevalent are p.Y129Y,p.H215R, p.P237S, p.I642M, p.I858V, p.N931N, andp.R1266Q.

The NPC2 gene, mutated in 5% of NPC families andmapped to 14q24.3, encodes a protein previously knownas HE1 [9]. This protein is a small (132 aa), soluble,ubiquitously expressed, and lysosomal glycoprotein thatspecifically binds unesterified cholesterol with high affin-ity [31–37]. At least 19 families with NPC2 mutationshave been identified to date ([9,26,38–40], this study).A total of 15 different mutations were identified, 11 ofwhich in a homozygous state. Molecular studies inNPC2 patients have so far shown very good correlationsbetween the genotype and the clinical phenotype but, byconstrast with NPC1, mutational studies gave littleinformation on NPC2 functional domains [40].

In medical practice, mutational analysis of patientswithNPC is important for propermanagement of affectedfamilies. Correlations have been found between geno-types and the severity of the neurological outcome ofthe patients [20,21]. Above all, molecular genetics consti-tutes the best approach for prenatal diagnosis [41].Molec-ular analysis of NPC patients is however challengingowing to the presence of two disease-causing genes, thesize of the NPC1 gene, the large number of private muta-tions and the occurrence of numerous polymorphisms.

Previous screening strategies have included SSCP,CSGE,and complete sequencing of eitherNPC1 cDNAorNPC1gene. These conventional methods for large-scale detec-tion of mutations are expensive and technically time-con-suming.DHPLChas been successful in overcomingmanyof these limitations and constitutes a detection methodwith a nearly 100% detection. This approach has allowedthe simple, semi-automated, and cost-effective detectionof single-base substitutions and small insertions/deletions.

Here, we report an optimized protocol for screeningthe NPC1 and NPC2 genes by DHPLC analysis andnovel mutations identified in 35 unrelated NPC patients,including 23 French families.

Materials and methods

Patients and biological material

For optimization of the DHPLC mutation detectionsystem, genomic DNA samples from 38 NPC patientswith identified genetic defects in either the NPC1 orthe NPC2 genes were used as positive controls. Themutations had previously been identified in our labora-tory by SSCP and sequencing analysis [15,20,38,40].

The optimized DHPLC conditions were applied toscreen mutations in a panel of 35 additional unrelatedNPC patients. The diagnosis had been established byevaluation of cellular cholesterol with filipin stainingand study of LDL-induced cholesteryl ester formationin cultured skin fibroblasts as described [5]. Classifica-tion of patients with respect to their clinical and bio-chemical characteristics was as proposed by us [2,5].In summary, patients with neurological symptoms werecategorized by age at onset of first neurological symp-toms, as having either a severe infantile form (onsetbefore 2 years of age), a late infantile form (onset atage 3–5 years), a juvenile form (onset between 5 and16 years) or an adult form (onset at age >16 years).Patients were also classified into either a classic or avariant biochemical phenotype [5]. The classic pheno-type refers to patients whose fibroblasts show a strik-ing cholesterol accumulation and a severe block inLDL-induced cholesteryl ester formation. In cells frombiochemically variant patients, demonstration of cho-lesterol storage requires challenge with pure LDL,and the rate of cholesteryl ester formation may notbe affected. Among the 35 patients studied, 27 showeda ‘‘classic’’ filipin staining pattern and 7 a ‘‘variant,’’although non-ambiguously positive, pattern. In onecase with a quite typical clinical history (case 22, code96092), the filipin staining pattern was only slightlyabnormal and it was felt difficult to definitively assessthe diagnosis of NPC from the cell biology/biochemis-try criteria.

Table 1

Conditions for amplification and DHPLC analysis of the NPC1 gene

NPC1

exon

Enzyme

usedaMgSO4

(mM)

T (�C) PCR product

length (bp)

Primer forward (5 0 fi 3 0) Primer reverse (5 0 fi 3 0) DHPLC analysis

Gradient

(%B)

T

(�C)Gradient

(%B)

T (�C) Gradient

(%B)

T

(�C)Gradient

(%B)

T

(�C)

1 FS ND 58 207 CCGCCGGCGTCAGCAGC TCGCCAGACCAACTTCCCCAGG 53.1 68 53.1 69

2 O 1.5 56 207 ACCATTGAGACCCTGGTAAC CATTTTGTGTTCCCAGTGCC 53.1 57.6 53.1 58.6 53.1 59.6

3 O 1.5 56 206 GACCTTACTCTAACTGTTGCC CACAAGTATCTACAGCCCAG 53.1 58.7 53.1 59.4

4 O 1.5 56 292 CTTGCTGGCCCTATTATGTGTG CAATTTGCTCTGCTGTCCTG 56.5 56.3 56.5 56.6 56.5 57

5 O 1.5 64 318 TGCCTCGTGAATTACAGCAAGCA CCCCAAGCACTGGTGAGCCACTG 57.2 58.2 57.2 59 57.2 59.5 55.2 61.5

6 O 1.5 56 551 ATTCCATAGGACGAAGCAGC CCATGCAATGGTATTCATGGAGG 61.2 54 58.7 56

7 O 1.5 56 285 GAAGGCAGTAATTAGGGAGG TGCAACCCACTGAGGAAACG 56.3 57 56.3 58.1 53.3 59

8 O 1.5 58 530 GTTCCGACTTTCAGGAACGGC AGCCCCAAATCCCCATCTAGC 57.4 61.9 56.7 62.9 56.2 63.9

9 O 1.5 62 334 TGACGTGTTTCTGGGTTTGCTTA TGCCCATGTACCCTAAGTCAGAC 56.1 57.5 55.6 59.2 55.1 60

10 O 1.5 52 286 AGGGCCCATGTTGTCCTTAGTGA TGCTAATGACAAAACCGAG 53.8 58.9 52.8 59.9

11 O 1.5 62 224 TGATTTTTCCCCTGGTATGTGTC CCCACAATGCAAGGACAGTCTG 53.9 57.1 53.9 58.1 50.4 59.1

12 O 1.5 58 343 CGTGGCCTTTGTATCGTGAAA TGACGTTACACTGTGCACTGCTG 54.9 57.4 54.4 58.4 53.9 59

13 O 1.5 64 311 AACAAGTGGGACAGACAACCCTG GCCCAGGAGCCATTCACAGTC 57 60.5 55 61.5 54.5 62.5

14 O 1.5 54 191 GTCGCATAATTTTTTTTTTTTTTT CCAGGCTCAGCAGACTACAGGAC 52.2 57.5 52.2 58.2 52.2 58.7

15 O 1.5 62 263 GAACATAAGACCTGCAGAGAGC CCGCTAGCTGCTTCCTCTAG 55.5 58.2 55.5 59.2

16 O 1.5 58 279 CTAGAGGAAGCAGCTAGCGG TCCTTCCCAGGCTGTCTGGC 56.1 58.5 56.1 58.9 56.1 59.6

17 O 1.5 54 209 TCCTGCTTTTTGTGTGTGCTTAA AAAAAAAAAAAAAAGGAAGTCATC 53.2 55.7 53.2 56.7 51.2 57.7

18 O 1 58 356 CTTATTCTCCGTGATCCTCGC CAGTGAGACATTTCAGGCCTG 58.2 57.5 55 61 54.5 62 54.5 62.5

19 O 1.5 58 257 AGACTTCCTCCCTGTGGAGC GGTATAAACTGAGGCACGATGC 55.3 56.9 55.3 58.9 52.1 59.4 51.8 59.9

20 O 1.5 58 303 GTAATGCCCCTCACTGTCAG GTCTTAGCCCAGTCCTCTCC 53.8 60.5 53.3 61.5 53.3 62

21 O 1 60 379 AATGTACAGCTGGGTCTGACC CAGTGTAGGCCCTTTGCTGG 58.6 59.7 58.6 59.9 58.6 60.2

22 O 1.5 62 330 CGGGAGTGAGAGCGAGCTTTAAT CCCCCAAGTGAAACAGGAGCTAG 57.5 60.3 58.2 60.7 57.5 61.3

23 O 1 58 234 AGCACCCATCCTCAGAACGG GTGCGACTCTGCCGGCGTGG 54.4 63.2 51.4 64.2 51.2 64.7 50.7 65.5

24 O 1.5 62 305 GGGGCAGGAGAATCACTTGAAC TCCATTGTGCCACCCTTTTAAGA 56.9 56.2 56.9 56.7 56.9 57.2

25 O 1 56 185 TGAGCCACTATGCCCAGCCAA GACACAGTTCAGTCAGGATG 53.4 58.7 52.9 59.7

a O, optimase (Transgenomic, Cheshire, UK); FS, fail safe mix enzyme (Epicentre, Madison, WI).

222G.Milla

tet

al./Molecu

larGenetics

andMeta

bolism

86(2005)220–232

Tab

le2

ConditionsforAmplificationan

dDHPLC

analysisoftheNPC2gene

NPC2

exon

Enzyme

useda

MgS

O4

(mM)

T(�C)

PCR

product

length(bp)

Primer

forw

ard(5

0fi

30 )

Primer

reverse(5

0fi

30 )

DHPLC

analysis

Gradient

(%B)

T(�C)

Gradient

(%B)

T(�C)

Gradient

(%B)

T(�C)

1FS

ND

58234

GGGCGGCCGCTGCTTCTTTC

GGCCGCCCGAGGGATCC

54.4

65.8

54.4

66.8

2O

1.5

60291

CTAAATGGGAGAGCAGAGCAC

CCCTCCATTCCCATGCTTATTC

56.5

58.2

56.5

58.7

53.5

61.5

3O

1.5

60477

ATGCTTGCACTGTGCTAG

TAC

TGATGCCTAACACCGCACCTA

60.3

5660.3

5860.3

594

O1.5

5826

6GGAGTTAGGAAGTTTGCTACTG

TGGCACTGATTTAGTTTCAGTC

55.7

57.4

55.7

58.4

55.7

59.4

5O

1.5

60199

TCGGAGTAAGGAAGGTCCAGCCA

TCAGCAGACTCATTGGCCAGAT

52.7

58.8

52.7

59.3

52.7

59.8

aO,optimase(Transgenomic,Cheshire,

UK);FS,failsafe

mix

enzyme(Epicentre,

Mad

ison,WI).

G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232 223

Amplification of NPC1 and NPC2 exons

Genomic DNA was isolated from cultured skinfibroblasts or from whole blood according to Jeanpi-erre [42]. All exons were amplified separately usingthe intronic primers and PCR conditions reported inTable 1 for NPC1 and in Table 2 for NPC2. For allexons except exon 1-NPC1 and exon 1-NPC2, PCRamplifications were performed in a final volume of50 lL containing 150 ng of genomic DNA, 1· PCRbuffer, x lL MgSO4, 200 lM of each dNTPs, 400 lMof each primer, and 1.25 U of Optimase polymerase(Transgenomic, Cheshire, UK) with the following cy-cling profile: 5 min denaturation at 92 �C, and 40 cyclesof denaturation at 92 �C for 45 s , annealing at theindicated temperature (Tables 1 and 2) for 1 min,extension at 72 �C for 1 min, followed by 5 min finalextension step at 72 �C. Exon 1-NPC1 and Exon 1-NPC2 were PCR-amplified in a final volume of 25 lLcontaining 75 ng of genomic DNA, 12.5 lL Buffer K(exon 1-NPC1) or G (exon1-NPC2), 400 lM of eachprimer and 1.25 U of FailSafe PCR Enzyme Mix (Epi-centre, Madison, WI) with the cycling profile describedabove. Prior to DHPLC analysis, the quality and quan-tity of PCR products were determined on 1.5% agarosegels by standard procedures.

DHPLC conditions

DHPLC analysis was performed using the WAVEDNA Fragment Analysis System (Transgenomic,Cheshire, UK). Prior to DHPLC analysis, PCR prod-ucts were denatured at 95 �C for 5 min and cooleddown at room temperature for 45 min. Samples(5 lL) were applied onto a preheated C18-reversedphase column of nonporous polystyrene-divinylbenzeneparticles (DNASep, Transgenomics). DNA was elutedwithin 2.5 min at a flow rate of 1.5 ml/min using a lin-ear acetonitrile gradient formed by mixing buffer A(0.1 M triethylamine acetate TEAA, pH 7.0) and bufferB (0.1 M TEAA, pH 7.0 with 25% v/v acetonitrile; Bio-solve, HPLC grade) at the respective denaturing tem-perature. The temperature for heteroduplex detectionwas determined using the WAVEMAKER software(Transgenomic, Cheshire, UK). For each exon, the pre-dicted temperature was optimized by testing controlDNA and, whenever available, DNAs carryingsequences variations on either NPC1 or NPC2 genes.Because some fragments showed distinct melting do-mains, additional analyzing temperatures were re-quired. After each run, regeneration of the columnwas achieved by washing with 100% buffer B(0.5 min) followed by an equilibration time of 2 min.The optimized elution profiles and melting tempera-tures of the entire coding sequences of NPC1 andNPC2 are given in Tables 1 and 2.

224 G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232

Sequence analysis

Any abnormal heteroduplex pattern obtained atDHPLC analysis was reamplified and subjected to directsequencing. PCR fragments were purified with the QIA-quick PCR purification kit (Qiagen, Valencia, CA).Sequencing reactions were performed using the CEQ2000 DTCS with Quick Start Kit (Beckman–Coulter, Fullerton, CA), purified by ethanol precipita-tion and applied onto an automatic sequencer (CEQ2000, Beckman–Coulter).

Results

Optimization of DHPLC conditions

Complete NPC mutational screening by DHPLC re-quired the investigation of 30 exons: 25 for the NPC1

gene and 5 for the NPC2 gene. For 5 of them (exons1, 3, 7, and 11 on the NPC1 gene and exon 5 on theNPC2 gene), no DNA carrying genetic variations wasavailable. For the 25 other exons, optimization ofDHPLC conditions was done using control DNA aswell as DNAs carrying known NPC1 or NPC2 geneticvariations (mutations or polymorphisms) previouslyreported by us [20,38] or that we identified later. Therunning temperatures selected for each fragment arereported in Tables 1 and 2. Up to 4 analysis tempera-tures could be required, due to the complexity of themelting domains for some fragments. To test our opti-mized conditions, a preliminary blinded study was per-formed using DNAs from 38 NPC patients previouslygenotyped by us (30 NPC1 patients and 8 NPC2patients) [20,38]. Without exception, all known dis-ease-causing mutations (n = 43) and single nucleotidepolymorphisms (n = 8) previously detected by SSCPwere readily identified as abnormal DHPLC profiles.Additional intronic or exonic polymorphisms notdetected by SSCP analysis were further identified(Table 3). Distinct DHPLC profiles were obtained foreach DNA variant. In the case of exons carrying both

Table 3Novel SNPs identified on the NPC1 gene with DHPLC analysis

Exon/intron SNP

2 p.S22SIVS4 IVS4 + 21 T > G9 p.T511MIVS11 IVS11 + 19 T > C18 p. P887PIVS12 IVS12 + 14 T > GIVS12 IVS12 + 16 delGIVS19 IVS19 + 28 C > T21 p.G1073G23 p.A1187A24 p.L1244L

a common polymorphism and a disease-causing muta-tion, DHPLC profiles differed from those observed forexons carrying only the disease-causing mutation orthe polymorphism. Detection efficiency and reproduct-ibility were evaluated and validated by performing 3separate amplifications for each exon.

Molecular screening strategy of NPC patients

Using the described DHPLC conditions (Tables 1and 2), molecular screening was performed in 33 newlydiagnosed NPC cases and in 2 cases in whom SSCPanalysis had allowed identification of only one mutat-ed allele (Table 4). To optimize genotyping, a strategyof screening was designed using the algorithm de-scribed in Fig. 1. Once biochemical diagnosis was as-sessed, DHPLC analysis was initiated by testing forthe presence of the frequent p.I1061T NPC1 mutation(exon 21). All cases with a biochemical ‘‘variant’’ phe-notype were also investigated for the two recurrentNPC1 mutations known to be associated with thisform, namely p.P1007A and p.G992R/W (exon 20).Except in cases in which this study led to the identifi-cation of the two mutated alleles, DHPLC analysis ofthe remaining NPC1 and NPC2 exons was processedfurther, giving priority to exons in which the largestnumber of mutations had so far been reported (exons6, 8, 9, 13, 18, 19, 22, and 24). Whenever DHPLCanalysis of all NPC exons did not allow identificationof the two mutated alleles, PCR products of exonscontaining polymorphisms were mixed in a 1:1 ratiowith another amplicon containing only this polymor-phism (Fig. 2). If the elution profile remained un-changed, the amplicon tested contained only thepolymorphism. If the elution profile differed from theoriginal one, there was indication that an additionalgenetic variation was present. In case of an NPC pa-tient with a known or suspected consanguinity, eachamplicon was systematically mixed in a 1:1 ratio withthat from a control patient as DHPLC analysis doesnot allow an easy detection of mutations in a homozy-gous status (Fig. 3). This was also done systematicallyfor the study of exon 21, due to the relatively high fre-quency of p.I1061T homozygosity.

Identification of novel NPC1 or NPC2 mutations andgenotype–phenotype correlations

Spectrum of NPC mutations

DHPLC analysis of 35 unrelated NPC patientsincluding 23 cases of French extraction allowed identi-fication of 45 disease-causing mutations of the NPC1gene in 33 of them (Table 4, Fig. 4), and of twoNPC2 mutations, p.P120S and p.Q146X, in two pa-tients with a consanguineous background. Geneticcomplementation analysis using cell hybridization con-

Table 4Clinical, biochemical, and molecular survey of 35 Niemann–Pick C disease patients

Patient code (origin) Clinicalphenotypea

Age atdeath (y)

Age at lastfollow-up (years)

Biochemicalphenotypeb

NPC1exon

NPC2exon

Genotype Effect on protein Reference

1 99010 (B) I 2 512 C 20 c. 2972–2973 delAG p.Q991fs [25]

20 c. 2972–2973 delAG p.Q991fs2 21124 (F) I 4 4

12 C 8 c.1129delC p.P377fs Novel mutation19 c.2909insT p.S970fs Novel mutation

3 22032 (Pa) I 4 612 C IVS10 c.1654+1G > T IVS10+1G > T Novel mutation

24 c.3718G > C p.G1240R Novel mutation4 93175 (FCa) I 6 C 10 c.1628C > T p.P543L [28]

10 c.1628C > T p.P543L5 21133 (F) I 5 3

12 C 4 c.423 delGAAA p.K142fs Novel mutation17 c.2594C > T p.S865L [30]

6 82042 (F) LI 5 912 C 8 c.1030insG p.S344fs Novel mutation

23 c.3557G > A p.R1186H [8]7 23050 (F/Pt) LI 6 C 5 c.530G > A p.C177Y [21]

? ? ?8 94097 (F) LI 9 C 5 c. 577G > A pW189X Novel mutation

18 c. 2749G > T p.D917Y Novel mutation9 96127 (F) LI 8 C 21 c.3182T > C p.I1061T [15]

22 c.3461T > A p.L1154X Novel mutation10 23015 (F) LI 8 C 12 c.1892T > G p.M631R [20,28]

21 c.3182T > C p.I1061T [15]11 94102 (F) LI 11 C 21 c.3182T > C p.I1061T [15]

21 c.3107C > T p.T1036M [8]12 20019 (F) J 6 C 8 c.1261C > T p.Q421X Novel mutation

21 c.3182T > C p.I1061T [15]13 250 (F) J 16 6

12 C 21 c.3182T > C p.I1061T [15]21 c.3185C > T p.A1062V Novel mutation

14 85091 (F) J 27 C 21 c.3182T > C p.I1061T [15]21 c.3182T > C p.I1061T

15 96054 (F) J 9 C 8 c.1210G > C p.R404P Novel mutation21 c.3182T > C p.I1061T [15]

16 98033 (F) J 15 C IVS3 c.288-12 delCTTTTC IVS3-12 delCTTTTC Novel mutation24 c.3647C > T p.A1216V Novel mutation

17 24129 (B) J 14 C 21 c.3182T > C p.I1061T [15]21 c.3182T > C p.I1061T

18 96091 (F) J 14 C 21 c.3182T > C p.I1061T [20]19 c.2819C > T p.S940L [14]

19 84045 (F) J 12 C 5 c. 497C > T p.P166L Novel mutation21 c.3182T > C p.I1061T [15]

20 24023 (F) J 25 C 12 c.1844G > T p.R615L Novel mutation? ? ?

21 82052 (B)c J 19 V 1 g.3096del31bp p.Q19fs Novel mutation17 c.2585A > T p.Q862L [20]

22 96092 (F) J 15 V(�)d 18 c. 2749G > T p.D917Y Novel mutation20 c.3019C > G p.P1007A [20]

(continued on next page)

G.Milla

tet

al./Molecu

larGenetics

andMeta

bolism

86(2005)220–232

225

Table 4 (continued)

Patient code (origin) Clinicalphenotypea

Age atdeath (y)

Age at lastfollow-up (years)

Biochemicalphenotypeb

NPC1exon

NPC2exon

Genotype Effect on protein Reference

23 98142 (G) J 22 V 20 c.3019C > G p.P1007A [20]21 c.3182T > C p.I1061T [15]

24 90135 (F) J 15 V 19 c.2903A > G p.N968S [29]24 c.3689insAATA p.L1230fs Novel mutation

25 22037 (F) J 18 V 10 c.1611T > A p.F537L Novel mutation20 c.3019C > G p.P1007A [20]

26 98137 (Al) A 29 C 15 c.2287T > C p.F763L Novel mutation18 c.2612A > G p.Y871C Novel mutation

27 96157 (G) A 37 C 16 c.2474A > G p.Y825C [20]20 c.2974G > A p.G992R

28 22108 (F) A 44 C 6 c.848 insGG p.F283fs Novel mutation22 c.3389G > A p.D1097N Novel mutation

29 96133 (F) A 25 V 19 c.2848G > A p.V950M [20]21 c.3182T > C p.I1061T [15]

30 21111 (F) A 21 V 20 c.2975G > A p.G992A Novel mutation20 c.2978 delG p.G993fs [30]

31 21121 (F) ? 3 712 C IVS22 c.3477+1 G > A IVS22+1G > A Novel mutation

23 c.3521C > T p.A1174V Novel mutation32 23090 (UK) ? 3 C 19 c.2801G > A p.R934Q [14,20]

20 c.2973 insA p.D994fs Novel mutation33 89090 (UK) ? 5 V 20 c.2974G > T p.G992W [13]

21 c.3182T > C p.I1061T [15]34 24084 (Tu) J 18 C 3 c.358C > T p.P120S* Novel NPC2 mutation

3 c.358C > T p.P120S*

35 25046 (Al) ? 0.5 C 4 c.436C > T p.Q146X Novel NPC2 mutation4 c.436C > T p.Q146X

Origin: Al, Algeria; B, Belgium; F, France; G, Germany; Pa, Pakistan; T, Turkey; Fca, French Canadian; F/Pt: French/Portuguese.a Classification of clinical phenotypes by age at onset of neurological symptoms, see Materials and methods. I, infantile; LI, late infantile; J, juvenile; A, adult; ?, yet unknown.b Defined by the degree of severity of alterations of intracellular cholesterol processing, see Material and methods; C, classic; V, variant.c Case 1 in [45,46].d See Material and methods.

226G.Milla

tet

al./Molecu

larGenetics

andMeta

bolism

86(2005)220–232

Clinical signs of NPC

Biochemical Diagnosis (on cultured fibroblasts)Filipin staining, LDL-induced cholesteryl ester formation

Negative

DHPLC screening of exon 21 NPC1 (p.I1061T)

and sequence analysis if abnormal profile

non-NPCpatient

GenotypedNPC patient

DHPLC analysis of theremaining NPC1 and NPC2 exons followed by sequencingof exons with abnormal elutionprofile

Completegenotype

Incomplete genotype

Incompletegenotype

NPC patient

DHPLC screening of exon 20 NPC1 (p.P1007A and p.G992R/W/) and

eventual sequence analysis

Completegenotype

Completegenotype

1 da

y

6 da

ys

«Classic»«Variant» or suspicious

1 day

Extraction of genomic DNA for molecular analysis

10 d

ays

Fig. 1. Flowchart of DHPLC strategy for NPC diagnosis. NPC1 or NPC2 exons were analysed using PCR and DHPLC conditions are described inTables 1 and 2.

G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232 227

firmed that the two latter patients belonged to theNPC2 complementation group (data not shown).DHPLC analysis allowed identification of the secondmutated allele for two patients (patients 6 and 27)for whom only one mutated allele had previously beendetected by SSCP analysis [20]. Only two mutant al-leles (3%) remained unidentified (patients 7 and 20).Among the 45 NPC1 mutations, 18 were alreadyknown, either reported by us (n = 9) [15,20] or by oth-ers (n = 9). [8,13,14,21,28,29]. Twenty-eight mutationsare reported here for the first time (8 frameshift; 3nonsense; 3 splicing defects, 14 missense mutations(Table 4). Most patients also carried several polymor-

phisms of the NPC1 gene and it was not uncommonthat a single NPC1 patient showed five genetic varia-tions of this gene or more. Since the exact functionof NPC1 is not known, expression studies are not reli-able for mutations that do not fully abolish cellularcholesterol transport—i.e., a majority of missensemutations, but because NPC1 polymorphisms havebeen well documented it is usually possible to definewhich changes are disease-causing.

When the mutational profile of the cohort studiedwas analysed globally, the largest number (n = 30) ofmutant alleles contained point mutations leading to sin-gle amino acid substitutions. Other genomic abnormali-

ig. 3. DHPLC elution profiles of p.I1061T identified in exon 21PC1 at the temperature of 60.2 �C. Pattern a, control patient; pattern, patient homozygous for the p.I1061T mutation; pattern c, patienteterozygous for the p.I1061T mutation; and pattern d, mix in a 1:1tio of an amplicon obtained from a control patient and of anmplicon obtained from a patient homozygous for the p.I1061Tutation.

Fig. 2. DHPLC elution profiles of two genetic variations (p.R934Q,IVS19 + 28 C > T) identified in exon 19 NPC1 at the temperature of59.4 �C. Pattern a, control patient; pattern b, patient heterozygous forthe p.R934Q mutation; pattern c, patient heterozygous for theIVS19 + 28 C > T polymorphism; and pattern d, patient heterozygousboth for the p.R934Q mutation and for the IVS19 + 28 C > Tpolymorphism.

228 G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232

ties included 3 nonsense mutations, 10 frameshift muta-tions, and 3 splice mutations. NPC1 mutations affectedvarious domains of the protein although they weremainly concentrated (n = 20) in the cysteine-rich lumi-nal loop (Fig. 4). The most common NPC1 mutation,p.I1061T, constituted 22% of the alleles, whilep.P1007A, the second most recurrent mutation, wasidentified in three alleles (19% of the alleles in the 8 ‘‘var-iant’’ patients). A hot spot region was observed aroundamino acid-992 where six different mutations were locat-ed (p.G991fs, p.G992A, pG992R, p.G992W, p.G993fs,and p.D994fs). Three missense mutations (p.R615L,p.M631R, and p.F763L) were located in the sterol sens-ing domain.

Correlations genotype–phenotypes

Among the eight patients with the less common var-iant biochemical phenotype, 4 were found to be com-pound heterozygotes for one previously describedvariant mutant allele: 3 patients with a p.P1007A allele(cases 22, 23, and 25) and one with a p.V950M allele(case 29). Three additional mutant alleles appeared tounderly a variant phenotype: p.Q862L (case 21),p.N968S (case 24), and p.G992A (case 30), since theywere found in combination with the severe mutationsp.Q19fs, p.L1230fs, and p.G993fs, respectively. Thenew ‘‘variant’’ alleles are all located in the cysteine-rich

FNbhraam

luminal loop. Surprisingly, cells from one patient with ap.G992R allele (case 27), so far always associated with avariant phenotype [20], showed an unambiguous classicbiochemical phenotype.

Correlations with the clinical phenotype could bemore easily discussed for patients carrying eitherhomozygous mutations (cases 1, 4, 14, 17, and 34)or mutations already defined phenotypically. Patients1 and 2 with frameshift mutations on both alleleshad the most severe neurological disease, with an earlyonset and a short life span, patients 14 and 17 homo-zygous for p.I1061T mutation had a juvenile neuro-logical onset form, while patients heteroalllelic forthis mutation (cases 9, 10, 11, 12, 13, 15, 18, 19, 23,and 29) presented different clinical phenotypes butnever showed the severe infantile neurological onsetform. Three missense mutations (p.P543L, p.S865L,and p.G1240R) appeared to be severe as they wereassociated to an infantile neurological phenotype(cases 4, 5, and 3). The present series comprised oneNPC2 patient homozygous for the missense p.P120Smutation located in the ECR D domain of theNPC2 protein. This patient had a ‘‘typical NPC’’ juve-nile neurological phenotype, rather uncommon forNPC2 patients. In patients 31–33 and 35, follow-up

Fig. 4. Topology of the mutations identified in 33 patients on an NPC1 protein model and correlation with the biochemical phenotype. Eachmutation was associated with a particular (‘‘classic’’ or ‘‘variant’’) biochemical phenotype after considering the contribution of the two NPC alleles.The schematic NPC1 protein model is drawn as proposed by Davis and Ioannou [12]. The blue areas indicate the putative transmembrane domains.The yellow oval frame delimits the SSD.

G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232 229

is too short to draw conclusions regarding their neu-rological subtyping. Patient 35 with the nonsensep.Q146X NPC2 mutation presented with a severeand progressive neonatal hepatic form. A paternalaunt with a diagnosis of NPC confirmed from thestudy of liver lipids had died at 2 months of age fromthe fatal neonatal cholestatic form.

Discussion

Mutation detection in NPC disease is a challenge be-cause NPC1, the disease causing gene in >95% of NPCpatients, is large in size (�50 kb) and mutations scat-tered over 25 exons. Several strategies have been usedby different authors: SSCP analysis on the NPC1 gene[14,18,30] or on the cDNA [17,20,21]; CSGE analysison the NPC1 gene [22,26]; direct sequencing of the gene[24], the cDNA [19], or of both the gene and cDNA[23,25]. These conventional methods for large-scaledetection of mutations are expensive, technicallydemanding, or time consuming. DHPLC appeared espe-cially well suited to the purpose of studying NPC1 andNPC2 genes, since the PCR products length for the var-ious exons varied between 185 and 551 bp, with a major-ity around 200–300 bp. Taking advantage of ourprevious experience in genotyping NPC patients

[15,20,21,38] and availability in the laboratory ofnumerous DNA samples with known variations of theNPC1 and NPC2 genes, we have now developed andvalidated specific DHPLC conditions allowing analysisof all individual NPC1 or NPC2 exons and their corre-sponding intron/exon boundaries. Considering the factthat the discrepancy between actual and displayed tem-perature in the column oven can increase over a periodof months, the use of more than one temperature perexon provides a potentially higher detection rate. Also,the use of only one oven temperature per exon harborsthe risk that some mutations in the low-melting domainsmay be missed. Performing the analysis at all selectedtemperatures indicated in Tables 1 and 2, a total of 88injections are required for each DNA sample. This num-ber may increase due to the need to reevaluate a numberof unknown amplicons after dilution with known ampli-cons (either normal or containing known genetic varia-tions). Using DHPLC conditions described in Tables 1and 2, we have made the estimate that the completemutation identification (DNA isolation, PCR reagents,DHPLC injections and sequencing) of 10 patients couldtechnically be achieved in approximately 2 weeks with areagent cost of approximately 200€/patient (techniciantime and machine cost not included), assuming thatthe machine would be fully available for that purpose.Compared with other methods, DHPLC is faster, less

230 G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232

laborious and appears as a high-capacity low-cost muta-tion detection method. Furthermore, DHPLC appliedto the analysis of the NPC1 and NPC2 genes provedto have a high sensitivity and efficiency.

In the random panel of 35 NPC patients that we stud-ied, 68 of 70 mutated alleles were identified: twenty-eightwere new mutations and 18 had previously been report-ed either by us (n = 9) [15,20] or by others (n = 9) [8,13–15,20,21,28,29]. Only 2 NPC1 alleles (3%) remainedunidentified. Using DHPLC, a nearly 100% throughputhas been reported for other genes. It should however bediscussed that essentially all published studies on NPC1gene have resulted in unidentified alleles, or alleles thatwere very difficult to identify. There are several reasonsto explain that DHPLC may fail to identify some muta-tions. Like all PCR-based analysis, DHPLC is not ableto detect deletions encompassing the whole gene or anentire exon. In the case of NPC patients with only onemutated allele identified after DHPLC analysis, furtherstudies on the cDNA sequence could be helpful [25].Such studies may reveal the presence of abnormal tran-scripts due to either (i) large genomic DNA deletions,(ii) intronic mutations distant from the exon/intron bou-daries and thus undetectable by DHPLC analysis, or(iii) exonic genetic variations not leading to an aminoacid change affecting an exonic splicing enhancer (ESE).

Among the 68 mutated alleles, we have identified 30missense mutations (65%), 4 nonsense mutations, 10frameshift mutations, and 3 splice mutations. More thanone-third of these mutations (41%) are concentratedwithin the cysteine-rich luminal loop and six of them(p.G991fs, p.G992A, pG992R, p.G992W, p.G993fs,and p.D994fs) are clustered in a hot spot region aroundamino acid-992. In this cohort, p.I1061T, the most com-mon NPC1 mutation, constituted 22% of the mutant al-leles and p.P1007A, the second-most-recurrent allele,4.5%. These frequencies are in excellent agreement withprevious studies [4]. The p.M631R missense mutationidentified in patient 10 is of particular interest as it af-fects one of the signals involved in the targeting ofNPC1 to late endosomes [43]. Three missense mutations(p.P543L, p.S865L, and p.G1240R) appear to be severeas they lead to an infantile neurological phenotype.Interestingly, two of them (p.P543L and p.G1240R)are not located on any of the three functionally criticaldomains of the NPC1 protein (Fig. 4). Within this co-hort, two NPC2 patients were genotyped. One patientwas homozygous for the p.P120S mutation located inthe ECR D domain of the NPC2 domain, suggested tobe involved in cholesterol binding [33]. This patienthas a juvenile neurological phenotype and he is still aliveat the age of 18 years, at variance with the majority ofNPC2 patients [36]. Patient 35, homozygous for thep.Q146X, is also an interesting NPC2 case as this muta-tion leads to the synthesis of a protein lacking only theC-terminal tripeptide SHL considered as a moderate

indicator for peroxisomal localization. Further expres-sion studies of these 2 NPC2 mutations could providenew insights on the NPC2 protein.

We have earlier concluded that specific NPC1 muta-tions define the ‘‘variant’’ biochemical phenotype andshowed that most ‘‘variant’’ mutations are located onthe cysteine-rich luminal loop [20,21]. Similar data havebeen simultaneously found by others [22]. As expected,all new cases with a variant phenotype included in thepresent study have one missense mutation in the cysteinerich luminal loop [20]. There has been a controversy asto one single allele with a ‘‘variant’’ mutant allele con-stantly defined a ‘‘variant’’ phenotype [22,30]. Our cur-rent data provide further indication that there are rareexceptions. Cells from patient 27, carrying onep.G992R allele, had a classic phenotype, whereas in allprevious studies, mutations affecting the codon 992 werealways reported in association with a variant biochemi-cal phenotype [13,14,20,25]. A similar discrepancy oc-curred for patient 24. The genotype of this patient ledus to conclude that the mutation p.N968S, located inthe cysteine-rich luminal loop, was responsible for theobserved variant biochemical phenotype, but a recentpublication reports a classic biochemical phenotype ina Chinese patient carrying the same mutation (also asso-ciated with a frameshift mutation) [29]. Two other dis-crepancies were reported for p.C177Y and p.P1007A.These two missense mutations were described as ‘‘vari-ant’’ mutations [20,21], but one patient with thep.C177Y [30] and one patient with the p.P1007A [22]were reported with a classic biochemical phenotype.More detailed studies including the complete polymor-phism pattern and possibly other factors may explainthese rare exceptions.

Identification of mutations in patients with NPC ismore important than in most of the other lysosomalstorage disorders, for several reasons. Diagnosis of pa-tients by demonstration of cholesterol accumulation incultured fibroblasts remains the gold standard method,and provides clearcut data in a majority of cases. Never-theless, from our experience based on the study of morethan 600 patients, we know that interpretation can bevery difficult in some cell lines showing only minorabnormalities. In borderline cases, rapid and low-costidentification of NPC1 mutations could be very usefulto reach a definitive diagnosis. Also, a problem oftenencountered in a newly diagnosed case is the difficultyto predict the neurological outcome. Many children willbe diagnosed with NPC in the first months of life or ear-ly childhood, well before neurological onset [4].Although genotype–phenotype correlations are limited,it is our increasing experience that in NPC1, some de-gree of prediction is often possible. The present studyconfirms our previous conclusion that one p.I1061T al-lele is sufficient to exclude the most severe infantile neu-rological form. Frameshift or nonsense mutations, as

G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232 231

expected, but also missense mutations affecting the ste-rol sensing domain usually have a severe impact. Onthe other hand, association with a mutation leading toan adult onset form when in the homozygous state usu-ally results in a slowly progressive juvenile or early adultonset form. Screening of NPC2 mutations by DHPLC,much easier than the genetic complementation tests bycell hybridization, should also allow the detection ofan increasing number of patients with alterations of thisgene. With emerging therapies, distinction of NPC1 andNPC2 cases, but also analysis of the mutations involved,might be of particular importance. The most urgent andcrucial purpose to identify mutations in NPC disease,however, is to facilitate prenatal diagnosis. The conven-tional cell biology technique [44] is demanding and isonly carried out in few specialized laboratories. Prenataldiagnosis using the molecular approach offers manyadvantages, among which of being faster and applicableto ‘‘variant’’ families [41]. Mutational analysis of the in-dex case should thus be a priority for all couples plan-ning a further pregnancy. The DHPLC strategy whichwe have developed fulfills all the conditions—rapidity,low-cost, and high efficiency—required for a more sys-tematic detection of NPC genetic variants and should al-low a better management of NPC patients and theirfamilies.

Acknowledgments

The authors are grateful to the patients and families,as well as to all colleagues who, over many years, pro-vided them with biological samples and with invaluableclinical information. This work was supported byINSERM/ AFM/French Ministery of Research (Re-search Network on Rare Diseases, Contract 4MR32F)and by Vaincre les Maladies Lysosomales.

References

[1] M.C. Patterson, M.T. Vanier, K. Suzuki, J.A. Morris, E.D.Carstea, E.B. Neufeld, E.J. Blanchette-Mackie, P.G. Pentchev,Niemann–Pick disease type C: a lipid trafficking disorder, in: C.R.Scriver, A.L. Beaudet, W.S. Sly, D. Valle, B. Childs, K.W.Kinzler, B. Vogelstein (Eds.), eighth ed., The Metabolic andMolecular Bases of Inherited Disease, McGraw Hill, New York,2001, pp. 3611–3634.

[2] M.T. Vanier, K. Suzuki, Niemann–Pick diseases, in: H.W. Moser(Ed.),Neurodystrophies andNeurolipidoses,HandbookofClinicalNeurology, vol. 66, Elsevier Science, Amsterdam, 1996, pp. 133–162.

[3] M.T. Vanier, K. Suzuki, Recent advances in elucidating Nie-mann–Pick C disease, Brain Pathol. 8 (1998) 163–174.

[4] M.T. Vanier, G. Millat, Niemann–Pick disease type C, Clin.Genet. 64 (2003) 269–281.

[5] M.T. Vanier, C. Rodriguez-Lafrasse, R. Rousson, N. Gazzah,M.C. Juge, P.G. Pentchev, A. Revol, P. Louisot, Type CNiemann–Pick disease: spectrum of phenotypic variation in

disruption of intracellular LDL-derived cholesterol processing,Biochim. Biophys. Acta 1096 (1991) 328–337.

[6] S.J. Steinberg, C.P. Ward, A.H. Fensom, Complementationstudies in Niemann–Pick disease type C indicate the existence ofa second group, J. Med. Genet. 31 (1994) 317–320.

[7] M.T. Vanier, S. Duthel, C. Rodriguez-Lafrasse, P. Pentchev, E.D.Carstea, Genetic heterogeneity in Niemann–Pick C disease: astudy using somatic cell hybridization and linkage analysis, Am. J.Hum. Genet. 58 (1996) 118–125.

[8] E.D. Carstea, J.A. Morris, K.G. Coleman, S.K. Loftus, D. Zhang,C. Cummings, J. Gu, M.A. Rosenfeld, W.J. Pavan, D.B.Krizman, J. Nagle, M.H. Polymeropoulos, S.L. Sturley, Y.A.Ioannou, M.E. Higgins, M. Comly, A. Cooney, A. Brown, C.R.Kaneski, E.J. Blanchette-Mackie, N.K. Dwyer, E.B. Neufeld,T.Y. Chang, L. Liscum, D.A. Tagle, Niemann–Pick C1 diseasegene: homology to mediators of cholesterol homeostasis, Science277 (1997) 228–231.

[9] S. Naureckiene, D.E. Sleat, H. Lackland, A. Fensom, M.T.Vanier, R. Wattiaux, M. Jadot, P. Lobel, Identification of HE1 asthe second gene of Niemann–Pick C disease, Science 290 (2000)2298–2301.

[10] M.E. Higgins, J.P. Davies, F.W. Chen, Y.A. Ioannou, Niemann–Pick C1 is a late endosome-resident protein that transientlyassociates with lysosomes and the trans-Golgi network, Mol.Genet. Metab. 68 (1999) 1–13.

[11] E.B. Neufeld, M. Wastney, S. Patel, S. Suresh, A.M. Cooney,N.K. Dwyer, C.F. Roff, K. Ohno, J.A. Morris, E.D. Carstea, J.P.Incardona, J.F. Strauss III, M.T. Vanier, M.C. Patterson, R.O.Brady, P.G. Pentchev, E.J. Blanchette-Mackie, The Niemann–Pick C1 protein resides in a vesicular compartment linked toretrograde transport of multiple lysosomal cargo, J. Biol. Chem.274 (1999) 9627–9635.

[12] J.P. Davies, Y.A. Ioannou, Topological analysis of Niemann–PickC1 protein reveals that the membrane orientation of the putativesterol-sensing domain is identical to those of 3-hydroxy-3-meth-ylglutaryl-CoA reductase and sterol regulatory element bindingprotein cleavage-activating protein, J. Biol. Chem. 275 (2000)24367–24374.

[13] W.L. Greer, D.C. Riddell, T.L. Gillan, G.S. Girouard, S.M.Sparrow, D.M. Byers, M.J. Dobson, P.E. Neumann, The NovaScotia (type D) form of Niemann–Pick disease is caused by aG3097– >T transversion in NPC1, Am. J. Hum. Genet. 63 (1998)52–54.

[14] W.L. Greer, M.J. Dobson, G.S. Girouard, D.M. Byers, D.C.Riddell, P.E. Neumann, Mutations in NPC1 highlight a conservedNPC1-specific cysteine-rich domain, Am. J. Hum. Genet. 65(1999) 1252–1260.

[15] G. Millat, C. Marcais, M.A. Rafi, T. Yamamoto, J.A. Morris,P.G. Pentchev, K. Ohno, D.A. Wenger, M.T. Vanier, Niemann–Pick C1 disease: the I1061T substitution is a frequent mutant allelein patients of Western European descent and correlates with aclassic juvenile phenotype, Am. J. Hum. Genet. 65 (1999) 1321–1329.

[16] J.A. Morris, D. Zhang, K.G. Coleman, J. Nagle, P.G. Pentchev,E.D. Carstea, The genomic organization and polymorphismanalysis of the human Niemann–Pick C1 gene, Biochem. Biophys.Res. Commun. 261 (1999) 493–498.

[17] T. Yamamoto, E. Nanba, H. Ninomiya, K. Higaki, M. Taniguchi,H. Zhang, S. Akaboshi, Y. Watanabe, T. Takeshima, K. Inui, S.Okada, A. Tanaka, N. Sakuragawa, G. Millat, M.T. Vanier, J.A.Morris, P.G. Pentchev, K. Ohno, NPC1 gene mutations inJapanese patients with Niemann–Pick disease type C, Hum.Genet. 105 (1999) 10–16.

[18] T. Yamamoto, H. Ninomiya, M. Matsumoto, Y. Ohta, E. Nanba,Y. Tsutsumi, K. Yamakawa, G. Millat, M.T. Vanier, P.G.Pentchev, K. Ohno, Genotype–phenotype relationship of Nie-mann–Pick disease type C: a possible correlation between clinical

232 G. Millat et al. / Molecular Genetics and Metabolism 86 (2005) 220–232

onset and levels of NPC1 protein in isolated skin fibroblasts, J.Med. Genet. 37 (2000) 707–712.

[19] V. Meiner, S. Shpitzen, H. Mandel, A. Klar, Z. Ben Neriah, J.Zlotogora, M. Sagi, A. Lossos, R. Bargal, V. Sury, R. Carmi, E.Leitersdorf, M. Zeigler, Clinical–biochemical correlation inmolecularly characterized patients with Niemann–Pick type C,Genet. Med. 3 (2001) 343–348.

[20] G. Millat, C. Marcais, C. Tomasetto, K. Chikh, A.H. Fensom, K.Harzer, D.A. Wenger, K. Ohno, M.T. Vanier, Niemann–Pick C1disease: correlations between NPC1 mutations, levels of NPC1protein, and phenotypes emphasize the functional significance ofthe putative sterol-sensing domain and of the cysteine-rich luminalloop, Am. J. Hum. Genet 68 (2001) 1373–1385.

[21] I. Ribeiro, A. Marcao, O. Amaral, M.C. Sa Miranda, M.T.Vanier, G. Millat, Niemann–Pick type C disease: NPC1 mutationsassociated with severe and mild cellular cholesterol traffickingalterations, Hum.Genet. 109 (2001) 24–32.

[22] X. Sun, D.L. Marks, W.D. Park, C.L. Wheatley, V. Puri, J.F.O�Brien,D.L.Kraft, P.A.Lundquist,M.C.Patterson,R.E.Pagano,K. Snow, Niemann–Pick C variant detection by altered sphingo-lipid trafficking and correlation with mutations within a specificdomain of NPC1, Am. J. Hum. Genet 68 (2001) 1361–1372.

[23] P. Bauer, R. Knoblich, C. Bauer, U. Finckh, A. Hufen, J. Kropp,S. Braun, B. Kustermann-Kuhn, D. Schmidt, K. Harzer, A. Rolfs,NPC1: Complete genomic sequence, mutation analysis, andcharacterization of haplotypes, Hum.Mutat. 19 (2002) 30–38.

[24] W.E. Kaminski, H.H. Klunemann, B. Ibach, C. Aslanidis, H.E.Klein, G. Schmitz, Identification of novel mutations in the NPC1gene in German patients with Niemann–Pick C disease, J. Inherit.Metab Dis. 25 (2002) 385–389.

[25] P. Tarugi, G. Ballarini, B. Bembi, C. Battisti, S. Palmeri, F.Panzani, E. Di Leo, C. Martini, A. Federico, S. Calandra,Niemann–Pick type C disease: mutations of NPC1 gene andevidence of abnormal expression of some mutant alleles infibroblasts, J. Lipid Res. 43 (2002) 1908–1919.

[26] W.D. Park, J.F. O�Brien, P.A. Lundquist, D.L. Kraft, C.W.Vockley, P.S. Karnes, M.C. Patterson, K. Snow, Identification of58 novel mutations in Niemann–Pick disease type C: correlationwith biochemical phenotype and importance of PTC1-likedomains in NPC1, Hum. Mutat. 22 (2003) 313–325.

[27] E. Di Leo, F. Panico, P. Tarugi, C. Battisti, A. Federico, S.Calandra, A point mutation in the lariat branch point of intron 6of NPC1 as the cause of abnormal pre-mRNA splicing inNiemann–Pick type C disease, Hum. Mutat. 24 (2005) 440.

[28] M.P. Scott, Y.A. Ioannou, The NPC1 protein : structure impliesfunction, Biochim. Biophys. Acta 1685 (2004) 8–13.

[29] C.C. Yang, Y.N. Su, P.C. Chiou, M.J. Fietz, C.L. Yu, W.L. Hwu,M.J. Lee, Six novel NPC1 mutations in Chinese patients withNiemann–Pick type C disease, J. Neurol. Neurosurg. Psychiatry76 (2005) 592–595.

[30] E.M. Fernandez-Valero, A. Ballart, C. Iturriaga, M. Lluch, J.Macias, M.T. Vanier, M. Pineda, M.J. Coll, Identification of 25new mutations in 40 unrelated Spanish Niemann–Pick type Cpatients: genotype–phenotype correlations, Clin. Genet. 68 (2005)245–254.

[31] N. Okamura, S. Kiuchi, M. Tamba, T. Kashima, S. Hiramoto, T.Baba, F. Dacheux, J.L. Dacheux, Y. Sugita, Y.Z. Jin, A porcinehomolog of the major secretory protein of human epididymis,

HE1, specifically binds cholesterol, Biochim. Biophys. Acta 1438(1999) 377–387.

[32] N. Friedland, H.L. Liou, P. Lobel, A.M. Stock, Structure ofa cholesterol-binding protein deficient in Niemann–Pick typeC2 disease, Proc. Natl. Acad. Sci. USA 100 (2003) 2512–2517.

[33] D.C. Ko, J. Binkley, A. Sidow, M.P. Scott, The integrity of acholesterol-binding pocket in Niemann–Pick C2 protein is neces-sary to control lysosome cholesterol levels, Proc. Natl. Acad. Sci.USA 100 (2003) 2518–2525.

[34] T.S. Blom, M.D. Linder, K. Snow, H. Pihko, M.W. Hess, E.Jokitalo, V. Veckman, A.C. Syvanen, E. Ikonen, Defective endo-cytic trafficking of NPC1 andNPC2 underlying infantile Niemann–Pick type C disease, Hum. Mol. Genet. 12 (2003) 257–272.

[35] K. Chikh, S. Vey, C. Simonot, M.T. Vanier, G. Millat, Niemann–Pick type C disease: importance of N-glycosylation sites forfunction and cellular location of the NPC2 protein, Mol. Genet.Metab 83 (2004) 220–230.

[36] M.T. Vanier, G. Millat, Structure and function of the NPC2protein, Biochim. Biophys. Acta 1685 (2004) 14–21.

[37] M. Zhang, M. Sun, N.K. Dwyer, M.E. Comly, S.C. Patel, R.Sundaram, J.A. Hanover, E.J. Blanchette-Mackie, Differentialtrafficking of the Niemann–Pick C1 and 2 proteins highlightsdistinct roles in late endocytic lipid trafficking, Acta Paediatr.(Suppl. 92) (2003) 63–73.

[38] G. Millat, K. Chikh, S. Naureckiene, D.E. Sleat, A.H. Fensom, K.Higaki, M. Elleder, P. Lobel, M.T. Vanier, Niemann–Pick diseasetype C: spectrum of HE1 mutations and genotype/phenotypecorrelations in the NPC2 group, Am. J. Hum. Genet. 69 (2001)1013–1021.

[39] H.H. Klunemann, M. Elleder, W.E. Kaminski, K. Snow, J.M.Peyser, J.F. O�Brien, D. Munoz, G. Schmitz, H.E. Klein, W.W.Pendlebury, Frontal lobe atrophy due to a mutation in thecholesterol binding protein HE1/NPC2, Ann. Neurol. 52 (2002)743–749.

[40] K. Chikh, C. Rodriguez, S. Vey, M.T. Vanier, G. Millat,Niemann–Pick type C disease: subcellular location and functionalcharacterization of NPC2 proteins with naturally occuring mis-sense mutations, Hum. Mutat. 26 (2005) 20–28.

[41] M.T. Vanier, Prenatal Diagnosis of Niemann–Pick diseases typeA, B, C, Prenatal Diagnosis 22 (2002) 630–632.

[42] M. Jeanpierre, A rapid method for the purification of DNA fromblood, Nucleic Acids Res. (1987) 9611.

[43] C. Scott, M.E. Higgins, J.P. Davies, Y.A. Ioannou, Targeting ofNPC1 to late endosomes involves multiple signals, including oneresiding within the putative sterol-sensing domain, J. Biol. Chem.279 (2004) 48214–48223.

[44] M.T. Vanier, C. Rodriguez-Lafrasse, R. Rousson, C. Marcais, J.Boue, A. Choiset, M.F. Peyrat, C. Dumontel, M.C. Juge, P.Pentchev, A. Revol, P. Louisot, Prenatal diagnosis of Niemann–Pick type C disease: current strategy from an experience of 37pregnancies at risk, Am. J. Hum. Genet. 51 (1992) 111–122.

[45] J.J. Martin, A. Lowenthal, C. Ceuterick, M.T. Vanier, Juveniledystonic lipidosis (variant of Niemann–Pick disease type C), J.Neurol. Sci. 66 (1984) 33–45.

[46] M.T. Vanier, Lipid changes in Niemann–Pick disease type Cbrain: personal experience and review of the literature, Neuro-chem. Res. 24 (1999) 471–479.