Editor—Niemann-Pick disease type C (NP-C, MIM 257220) is a fatal autosomal recessive disorder characterised by progressive neurological deterioration and hepatosplenomegaly. NP-C patients can be classified into four major groups according to the onset of neurological symptoms, that is, early infantile, late infantile, juvenile, and adult forms, and the earlier the clinical onset the more quickly progressive are the symptoms and the shorter is the life span.1-4Complementation analysis using cultured skin fibroblasts indicated the presence of at least two subgroups of NP-C, NPC1 (the major subgroup that comprises >90% of NP-C patients) and NPC2 (the minor subgroup).2-4 In 1997, the NPC1 gene (NPC1) (accession No AF002020) that is responsible for the NPC1 subgroup was identified by positional cloning.5 6 The number ofNPC1 mutations known to date is not far off 100,7-11 taking into account the accumulated data from seven groups presented in a recent international workshop (International Workshop, The Niemann-Pick C Lesion and the Role of Intracellular Lipid Sorting in Human Disease, Bethesda, USA, October 1999).

Because the genomic structure of NPC1 was unknown, initial mutation screening was performed on RT-PCR products or partial genomic amplicons. In our previous study using RT-PCR products, we identified 14 different mutations in 19 alleles from 11 patients, and failed to detect mutations in the remaining three alleles.8 Mutation screening using RT-PCR products has several drawbacks compared with screening using genomic amplicons. For example, mutations that reduce the mRNA stability may escape the screening.12 13 To refine the screening method, we screened a CITB human BAC library (Research Genetics, Huntsville, AL) and isolated a clone 386K10 that contained all the 25 exons ofNPC1 and a 2 kb fragment of 5′UTR. Our analysis using 386K10 confirmed the exon/intron boundary sequences reported by Morriset al 14 and complements their data by showing the lengths of introns 1 (20 kb) and 6 (3 kb). Thus, NPC1spans over 70 kb in the genome.

Sets of primers to amplify each of the 25 exons ofNPC1 were designed according to the corresponding intron sequences (table 1). To include cis acting elements that participate in pre-mRNA splicing, the 3′ nucleotide of nearly all the primers was placed >20 bp away from the splice junctions. For SSCP, exons 6, 8, and 9 were divided into two to three fragments by primers based on each exon sequence and named exon 6a and 6b and so on (table 1). The clinical features of the 15 Japanese and two white NPC1 subjects are summarised in table 2. All the patients were diagnosed by cholesterol accumulation in their skin fibroblasts.15 Informed consent for gene research was obtained from all the families. Two NPC1 cell lines (GM03123 and GM110) were obtained from the Human Genetic Mutant Cell Depository, Coriell Institute for Medical Research (Camden, NJ). Fibroblasts from one healthy volunteer and three NPC2 patients were used as controls.

By SSCP analysis of genomic amplicons, we surveyed the 34 alleles from the 17 patients (including the 11 subjects in our previous study8), confirmed the 14 mutations that had been identified by RT-PCR SSCP, and identified one recurrent and seven novel mutations (table 3). None of the recurrent or the seven new mutations were found in over 100 normal samples, and they were thus considered to be disease causing. Mutation S954L identified in 431-1 is a recurrent mutation that has been reported by Greer et al 7 and also by Bauer et al(International Workshop, The Niemann-Pick C Lesion and the Role of Intracellular Lipid Sorting in Human Disease, Bethesda, USA, October 1999). Of the seven novel mutations, five were found in new subjects whereas the remaining two were found in one allele of TAN (C3614G) and of SAK (3615 (−3618) A del), respectively. It is not known why these two mutations escaped RT-PCR SSCP. Allelic mutations were not detected in three patients (OHS, SAS, and YAN) (table 3). In summary, SSCP analyses of genomic amplicons showed 21 disease causing mutations in 31 out of 34 alleles from 17 patients. Additionally, six different variants were identified (table 4).

The 22 mutations included 15 missense mutations, two nonsense mutations, two in frame deletions, and three deletions that cause a frameshift and a premature stop codon. In accordance with our identification of T3182C (I1061T substitution) as a frequent mutant allele in patients of western European descent,13 this mutation was found in the genome of two white cell lines. None of the Japanese patients possessed this mutant allele, clearly highlighting an ethnic difference in the mutation frequency. Instead of T3182C, G1553A appears to be a relatively frequent mutation in Japanese patients, found in five alleles in three patients. This mutation is unique for two reasons; one is that it is predicted to cause both an amino acid substitution (R518Q) and an alternative exon skipping8 and the other is that the skin fibroblasts from patients homozygous for this mutation (KUR and INO) retained considerable levels of NPC1 protein (see below).

With regard to the structure-function relationship ofNPC1, mutagenesis studies have shown several functionally important domains of NPC1 protein including an NPC domain and a sterol sensing domain (SSD).16 17 In addition, Greer et al 9 suggested the functional importance of the cysteine rich extracellular loop between TM9 and TM10 based on the segregation of point mutations in this region. The 14 missense mutations and the one in frame deletion found in the present survey are widely distributed on NPC1 cDNA and appeared to be classified into five groups according to their location (fig 1A). Each group of mutations gives some insight into the structure-function relationship of NPC1. First, two mutations (F703S and del 740-741) in group II are located in the sterol sensing domain. Second, four mutations in group VI are located in the cysteine rich extracellular loop. Interestingly, C956Y is the mutation of the cysteine residue itself that is supposed to be involved in the secondary structure formation and the other two mutations (V889M and M996R) were located in the conserved motif sequences in this loop (fig1B). Thus, the mutations in groups III and IV appear to reinforce the functional significance of SSD and the cysteine rich domain, respectively. By analogy, one may infer the presence of functionally important domains that correspond to groups I, II, and V mutations and this should be the subject of a future study. No wild type mutations were found in the NPC domain, although the functional importance of this domain is obvious from mutagenesis studies.17

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Distribution of mutations in NPC1. (A) Missense mutations and in frame deletions identified in this study are depicted. Circles and triangles indicate missense mutations and in frame deletions, respectively. Black circles and triangles are the mutations found in late infantile form patients and grey ones are from juvenile and adult form patients. Underlined are mutations found in white cell lines. (B) Mutations in NPC1 specific cysteine rich domain. Black circles are cysteines that may form disulphide bonds and grey ones are conserved amino acids. Squares are mutations reported by Greer et al.7 9 Underlined are mutations found in patients with moderate or mild phenotypes. The model for the organisation of NPC1 is according to Greer et al.9 Still tentative, it may have to be slightly altered in the future.19

To investigate the impact of mutations on expression of the translation product, we quantified the levels of NPC1 protein in membrane preparations from cultured fibroblasts by anti-NPC1 immunoblotting18 (fig 2). The anti-NPC1 detected two bands on the blot of the control membrane preparations, a major band at ∼170 kDa and a minor band at ∼190 kDa. These two bands have been shown to represent the same protein with differential glycosylation.16

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Western blot of membrane proteins extracted from skin fibroblasts of NP-C patients and normal controls (C). Numbers 1 to 3 indicate NPC2 patients. Molecular weight (kDa) is given on the left. A rabbit polyclonal anti-NPC1 antibody was a kind gift from Dr S C Patel and was used at 1:100.

In NPC1 cell lines, there appeared to be a distinct difference in the NPC1 protein levels between the late infantile and juvenile/adult forms. In the late infantile forms, there was a clear reduction of the NPC1 protein level regardless of the type of mutation, and five fibroblast lines (MUR, OHS, SHI, GM3123, and GM110) expressed undetectable levels of NPC1 protein. An exception was KUR and INO, both of whom have R518Q homozygous mutations and levels of NPC1 protein in their fibroblasts were close to those of controls.

Patients with a late clinical onset were distinct in that all of their skin fibroblasts expressed considerable levels of mutant NPC1 protein (fig 2). Two of the three patients (END and KAI) with a late clinical onset were compound heterozygotes for the groups IV and V mutations, whereas at least one allelic mutation of the 14 patients with a late infantile form belonged to group I, II, or III (fig 1A). In another study, skin fibroblasts from a patient with an adult neurological onset (homozygous for a V950M mutation)11appeared to retain normal expression of NPC1 protein (G Millat, M T Vanier, C Tomasetto, unpublished data). These results led us to form a tentative conclusion that the relatively mild form of NPC1 is caused by mutations located on the C-terminal side of the transcript that do not interfere with expression/turnover of the translation product. Future studies with an increased number of patients will verify this conclusion.

Finally, we also found that NPC2 fibroblasts expressed normal, or rather increased levels of NPC1. Similar results were achieved in a parallel study conducted with another antibody (G Millat, M T Vanier, C Tomasetto, unpublished data). Because of the identical biochemical phenotype of NPC1 and NPC2, the NPC2 protein is assumed to be located close to NPC1 both spatially and functionally. At one extreme, there has been a hypothesis that the biochemical phenotype of NPC2 is the result of the secondary absence of NPC1.2-4 Our findings clearly exclude this hypothesis but do not exclude that NPC2 is required for the normal function of NPC1.