Material and methods

Skin fibroblasts of MPS IIIB patients were obtained for enzyme diagnosis and stored at the European Human Cell Bank in Rotterdam (WJK) or at the University Children’s Hospital in Mainz (MB). Genomic DNA was isolated from cultured fibroblasts or whole blood according to standard procedures and total RNA was isolated from fibroblasts as described elsewhere.9 Polymerase chain reaction (PCR) of NAG exons including adjacent intronic regions was done with the primers listed in table 1. Exon 1 was amplified in two and exon 6 in five overlapping fragments. PCR reactions contained 20 mmol/l Tris-HCl (pH 8.4), 50 mmol/l KCl, 1.5 mmol/l MgCl₂, 0.2 mmol/l of each dNTP, 0.4 μmol/l of each forward and reverse primer, 1 unitTaq polymerase (Life Technologies), and 100 ng template DNA. For exon 1.1, 10% (v/v) DMSO was added. PCR was done with a five minute initial denaturation step, followed by 35 cycles of one minute at 94°C, one minute annealing, and 70 seconds extension at 72°C. Annealing temperature was 58°C except for exon 1.2 (60°C), exon 2 (62°C), exon 3 (50°C), and exon 6.2 (54°C). Non-radioactive screening for sequence alterations was done combining SSCP/heteroduplex analyses as previously described.9 Most fragments were electrophoresed on 8% polyacrylamide gels containing 1 × TBE and 10% glycerol at 22 W for 18 hours. Exon 1.1 and 1.2 fragments were separated on gels without glycerol for six hours at 27 W. For exon 6.3, 5% glycerol was used (17 W, 18 hours).

Fragments showing mobility shifts of single strands and/or heteroduplex formation were sequenced directly using the ABI Prism Dye Terminator Kit (PE Applied Biosystems) and an ABI Prism Sequencer (model 377, PE Applied Biosystems). In most cases, amplification primers were used as sequencing primers.

Reverse transcription of RNA was performed with random hexamer priming as described previously.9 The NAG coding region was amplified in three overlapping fragments using the primers listed in table 1. PCR conditions were identical to those used for genomic fragments, except that the reaction for c-NAG 1 contained 10% DMSO as well as 0.5 U of Perfect Match (Stratagene) and annealing temperature was 52°C. cDNA fragments were sequenced without previous screening procedures as described above.

All mutations were verified by restriction analysis or by sequencing a second, independent PCR product. One hundred control chromosomes were analysed for each mutation using SSCP, restriction analysis, or allele specific oligonucleotide (ASO) hybridisation. ASO analysis was performed as described previously10 and sequences of ASOs are available on request.

Results and discussion

All NAG gene exons were amplified in fragments suitable in size for SSCP analysis (<400 bp9). DNA fragments were analysed on SSCP gels which allowed inspection of double strands at the same time. Table 2 summarises the mutations/sequence alterations identified in 22 patients (note that nucleotides are numbered according to reference 5). As shown in table 2, clear single strand mobility shifts were detected for most of the mutations and were accompanied, in many cases, by heteroduplex formation. However, some mutations did not produce clear mobility shifts of single strands but were detectable by heteroduplex formation only. This was unexpected, especially for 1006delAG with two very prominent heteroduplex bands but with an apparently unaltered single strand pattern. One possible explanation for this would be that the fragment with the deletion has a great number of different conformations which do not produce any defined single stranded bands under the electrophoretic conditions used.

For those patients in whom either one or both disease alleles could not be identified by exon screening, sequencing of the entire cDNA was performed. In this way, five more point mutations were detected (table2) which were confirmed subsequently at genomic level. Even with this combined screening method, only 82% of disease alleles were identified, a proportion that is similar to that we obtained for genes involved in other lysosomal storage disorders, that is, the α-L-iduronidase gene (88%11), the N-acetyl-galactosamine-6-sulphatase gene (88.6%10), and the sulphamidase gene (86%12). It could be that some of the mutations are located in regulatory sequences of the NAG gene or in those parts of introns that were not analysed.

Of the 18 point mutations reported here, eight (G79C, R203X, G292R, R297X, E452L, R482W, R565Q and R674H) affect highly mutable CpG dinucleotides. In the latter seven cases, a C to T or G to A transition is present, being in line with methylation mediated deamination of 5-methylcytosine as one of the possible mechanisms of mutagenesis. Two of these mutations, R297X and R674H, have been found previously in most likely unrelated patients6 suggesting that the nucleotides involved are more prone to mutation.

It is worth mentioning that two mutations described here are located very close to mutations reported earlier. The insertion 338ins24 is situated only four nucleotides downstream of 233ins247(different numbering of nucleotides according to reference 6). As these insertions both involve a GC rich area with four perfect GGC and three imperfect GCG repeats, and as 388ins24 occurred in two probably unrelated patients, this region may be a “hot spot” for slipped mispairing replication errors. The deletion 1006delAG starts four nucleotides downstream of 901delAA8 (numbering of nucleotides according to reference 6). In both cases, the mutations may have quite similar molecular consequences.

One of the two more frequent mutations in our sample, Y140C, was identified on three alleles (6.8%). As the patients harbouring this mutation originate from Germany, Italy, and the Czech Republic, respectively, the gene alterations are probably of independent origin. Y140C was also identified in other patient groups.7 8R565Q was seen on 6.8% of the alleles (two German and one Turkish patients) and 338ins24 occurred in two patients from Iran and Turkey. All other mutations were found in one patient only. There was a high incidence (45%) of homozygosity. Consanguinity of parents was known in three cases (B7 from Turkey, B18 and B19 from Iran), but some other patients may also originate from consanguineous marriages, especially those from Iran and Turkey. However, we cannot exclude that some of the patients with apparent homozygosity for a given mutation are in fact heterozygous with another mutation which escaped PCR amplification, for example, owing to a larger deletion.

Two frameshift mutations (274ins4 and 1006delAG) and three nonsense mutations (R203X, R297X, and W404X) are likely to lead to instability of the mRNA. Moreover, any protein translated from these transcripts would be largely truncated with most probably no enzyme function. A duplication of eight amino acids (338ins24) as well as the deletion of one amino acid (F142del) may also have a major impact on the protein structure and function.

In exon 1, a point mutation was found in the initiation codon which changed methionine to leucine. Initiation codon mutations have been described occasionally and some were proven to abolish translation.13 A Met→Leu initiation codon mutation was also described for the sphingolipid activator protein gene and produced no detectable protein.14 As there is no other ATG in the vicinity of the initiation codon of the NAG gene, it seems that no alternative initiation of translation can take place.

The majority of the mutations identified here predict the replacement of only one amino acid (table 2). In each case, 100 control chromosomes were screened for the absence/presence of the missense mutations. Only G737R was found on two out of 100 control chromosomes in heterozygous state. Glycine at position 737 is not conserved between the human and mouse NAG genes, which have 82% identity at the amino acid level, the mice having glutamine at this position. It is worth noting that three of the four patients with G737R have R565Q. Family analyses will be necessary to prove if these two alterations are indeed located on one allele. G737R may be more frequent among patients (9%) than among controls (2%) because R565Q occurred on a G737R allele in a common ancestor.

All other missense mutations presented here were not detected on control chromosomes, affect amino acid residues conserved between human and mouse NAG, and were, therefore, considered most likely pathogenic. The only exception is R100H where the mouse protein bears glutamine. The facts that most missense mutations affect conserved residues and that most of them are non-conservative changes make their pathogenic nature probable, but enzyme function impairment should be proven by expression studies.

One polymorphic nucleotide exchange was detected in intron 2 (IVS2+50G→C) with allele frequencies of 0.67 (C) and 0.33 (G) in patients. Screening of 100 control chromosomes showed similar allele frequencies (0.7 and 0.3). The absence of linkage disequilibrium is not unexpected as no common disease allele was identified. The polymorphism may help to identify disease alleles in families in which the pathological mutations have not yet been detected. Carrier testing was done in the past by measuring enzyme activities, but a considerable overlap was present between obligate heterozygotes and controls.16 Identification of primary gene lesions now allows unambiguous carrier testing and will help in counselling the families concerned. This is especially important in populations with a tradition of consanguineous marriages (for example, families from Iran and Turkey in this study).

The more pronounced clinical heterogeneity in MPS IIIB compared to MPS IIIA2 is probably because of the genetic variability found. In our patient sample, nearly all affected subjects had a typical, severe course of the disease with symptoms present as early as 2 to 4 years of age and enzyme deficiency diagnosed, in most cases, between 2 and 8 years of age. Only one patient (B3, R565Q, other allele not identified) had an attenuated form of MPS IIIB; at the age of 15 she was still able to speak. In two patients, one from Germany (B9, genotype C277F/M1L) and the other from Iran (B19, genotype R100H/R100H), a very severe course of the disease was described with onset at birth. The fact that many of the missense mutations described here are associated with a severe clinical picture in homozygotes may help to pinpoint amino acids essential for enzyme function.

In summary, the vast genetic diversity seen in MPS IIIB was unexpected and unlike some other autosomal recessive inherited mucopolysaccharidoses. For MPS I, two very common mutations were found in the α-L-iduronidase gene11 15 and for MPS IIIA common mutations exist in the sulphamidase gene.12 17 In contrast, extensive genetic heterogeneity was found in MPS IVA,10 which resembles that emerging now for MPS IIIB. Other groups have previously identified 24 different mutations of the NAG gene, none of them common.5-8 The results presented here almost double the number of mutations identified for this gene.