Subjects and methods

A total of 57 families was investigated, including most known patients with GA1 in Germany. Most patients in the study had classical GA1 but some displayed atypical clinical or biochemical features, such as consistent isolated urinary excretion of 3-hydroxyglutaric acid, or were referred for exclusion of GA1. GDH deficiency was previously or subsequently proven by enzyme analysis in 46 patients (E Christensen, Copenhagen). Normal GDH activity was found in eight patients, while no enzymatic data were available in three patients. Patients and, where available, both parents were investigated; in one family no patient DNA was available. The families studied were unrelated, but the parents of some patients were related or came from areas with a high degree of consanguinity. The ethnic or national background of each parent over several generations was recorded. The origin of families with proven GDH deficiency was as follows: 18 German, nine Turkish (seven living in Germany), four Austrian, six Chilean, and 11 from different European countries and the United States.

Genomic DNA was extracted from peripheral lymphocytes using standard methods. Primer sequences for all exons are given in table 1. For DGGE analysis, a 40-65 bp GC clamp was attached to one of the two exon specific primers. PCR conditions were 95°C for three minutes; 35 cycles of 95°C for 30 seconds, 60°C for 30 seconds, 72°C for one minute; followed by 72°C for five minutes, 95°C for one minute, 65°C for 30 minutes, 37°C for 30 minutes, 4°C hold (Taq polymerase purchased from Life Technologies, Rockville, MD, USA; thermocycler 9700, Perkin Elmer, Norwalk, CT, USA). Amplification of exon 1 for DGGE required the addition of 5% DMSO. Broad range DGGE was performed on a vertical electrophoresis system (modified Protean II, BioRad, Hercules, CA, USA) using essentially the same procedure and electrophoresis conditions as previously described for the PAH gene.7 All exons were run in parallel on the same gel containing a denaturing gradient from 0-80%; electrophoresis time was six hours at 170 V. The relatively short PCR product of exon 8 was loaded 90 minutes after start of the electrophoresis run. The results were visualised by ethidium bromide staining. Exons displaying variant gel patterns were amplified from genomic DNA with primers containing a universal 20mer tail (table 1), cycle sequenced in both directions with dye labelled universal primers (ThermoSequenase, Amersham Pharmacia, Uppsala, Sweden), and electrophoresed on an ALF express fluorescent sequencer (Amersham Pharmacia, Uppsala, Sweden). Inheritance in the families was confirmed by DGGE or sequencing. Several mutations including R402W were confirmed by appropriate restriction enzyme analyses of the respective exon.

The inheritance of three polymorphic silent mutations in theGCDH gene was investigated with restriction enzyme assays within families. IVS2+63G>C creates anAccI site in intron 2, while c.1173G>T (G391G) in exon 10 and c.1482A>G in the 3′ untranslated region require amplification with specifically altered PCR primers (table 1), artificially creating RsaI andBsiEI restriction sites, respectively (ACRS method8). For the characterisation of alleles using these markers, a small letter (a-h) was assigned to each possible haplotype combination. The numbering of cDNA nucleotides (c.) follows the recommendations of the Nomenclature Working Group9 and starts with 1 at the first nucleotide of the initiator codon; values are thus 36 nucleotides smaller than those used in previous publications on GA1 mutations.

Results

The result of broad range DGGE analysis of all exons of theGCDH gene in four GA1 patients is illustrated in fig 1. Mutations were found on all 96 GA1 alleles in 48 patient families with unequivocal diagnosis of the disease, including two of the three patients in whom enzyme studies were not performed (table 2). There were 38 different mutations including 20 that have not been previously reported. All but one small deletion were single nucleotide changes; there were three splice site mutations. R402W was confirmed to be the commonest mutation in Europeans, accounting for one quarter of mutant alleles (24/96) in our study. Ten other mutations were also found in more than one family. Fourteen mutations, most of them previously described, were C>A or G>T alterations at potentially methylated CpG sites that are known to be hypermutable.10Origin of alleles and haplotype associations are included in table 2. Among both normal and mutant alleles, there were two major (a and h) and one minor (b) haplotype; the other haplotypes were rare or absent. There was no significant difference in the haplotype distribution between normal and mutant alleles (table 3). With regard to individual mutations, more than half were associated with haplotype a. No mutations were found in the nine patients with atypical clinical or biochemical findings in whom GDH enzyme analysis was normal (eight) or not available (one).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Broad range DGGE in four patients with GA1. Patient 1 is heterozygous for polymorphism IVS2+63G>C in intron 2, mutation R88C in exon 3, mutation A195T in exon 6, and polymorphism G391G in exon 10. Patient 2 is heterozygous for polymorphism IVS2+63G>C, mutation R227P in exon 7, and mutation R402W + polymorphism G391G (in cis) in exon 10. Patient 3 is heterozygous for novel silent mutation IVS2+48C>T in intron 2, mutation R138K in exon 5, and mutation L309W in exon 8. Patient 4 is heterozygous for polymorphism IVS2+63G>C, mutation R132Q in exon 5, mutation R355H in exon 9, and polymorphism G391G. Polymorphism c.1482A>G is not contained in the PCR product used for DGGE analysis of exon 11.

R402W was more common in patients of German origin in whom it had a relative frequency of almost 40% (14/36 alleles), but it was also identified in patients from various other European countries as well as Chile and the USA. The mutation is usually associated with haplotype h but was found once on a haplotype a background; as a CpG mutation it may have arisen independently in different founders. Of the four Austrian patients, three were heterozygous and one was homozygous for R88C, a mutation that was not identified in patients from other countries. This clearly reflects a strong regional founder effect. The common “Amish” mutation A421V6 was detected in patients from Germany, Austria, Croatia, and Italy, indicating a central European origin of the mutation in line with the known historical background of the Amish people. In 22 patients, mutations were found in homoallelic constellations; blood samples from both parents were available in 15 of these families. The father in one Chilean family did not carry the mutation (Y113H) for which his child was homozygous, reflecting either non-paternity or hemizygosity of Y113H with a large deletion on the paternal chromosome.11Five of the eight homozygous German patients carried R402W, including one who was heterozygous for GCDHhaplotypes. Seven of the nine Turkish patients were homozygous, and the comparatively high rate of consanguinity probably contributes to the relatively large number of Turkish patients with GA1. The commonest Turkish mutations with relative frequencies above 20% were P248L (5/18 alleles) and E365K (4/18 alleles). P248L was also identified in a compound heterozygous patient from Sicily, while the CpG mutation E365K was found in association with a different haplotype in a North American patient, again possibly reflecting independent recurrence. Differing haplotype associations were also found for one other CpG mutation, R132Q, in two patients of German and Turkish origin.

Discussion

GA1 is very heterogeneous, and the pattern of mutations is typical for a disease that does not confer heterozygous advantage. Most mutations are private to individual patients, and the few more common mutations probably achieved their higher relative frequencies through recurrence, founder effect, and genetic drift. There is one relatively common mutation, R402W, that may be partly responsible for the relatively high incidence of GA1 in European populations. A large number of mutations involve hypermutable CpG sites in the gene, as would be expected for a disease in which a strong negative selection pressure (homozygosity for GA1 alleles is usually lethal if untreated) is counterbalanced by new mutations. The high proportion of homozygosity and consanguinity among GA1 families indicates that the frequency of mutant alleles (carrier status) in the general population is probably smaller than would be calculated using the Hardy Weinberg formula.

Previous reports have suggested that compound heterozygosity for mutation R227P and another mutation in theGCDH gene is associated with lower urinary excretion of free glutarate,12 and this observation was confirmed in our study. All three patients who carried R227P on one allele had atypical biochemical findings in urinary organic acid analysis (data not shown), and one of them had unusual isolated 3-hydroxyglutaric aciduria.13 This is most probably the result of significant residual enzyme activity, as has been shown recently by expression analysis of R227P in E coli.14 However, no strict correlation between biochemical and clinical phenotypes has yet been reported. It is interesting to note that all three patients with R227P were of Anglo-Saxon origin, and it is possible that this mutation may be more prevalent in this population.

Our results indicate that mutation analysis is a useful alternative for confirmation of diagnosis in patients with suspected GA1 when enzyme analysis is not available or feasible. We identified mutations on both alleles in all patients with enzymatically proven GDH deficiency including several patients with unusual biochemical features such as isolated excretion of 3-hydroxyglutaric acid or normal urinary organic acid profile. No mutation was found in patients in whom the diagnosis of GA1 was excluded through normal enzyme studies. Among these patients there were two with clinical features compatible with GA1 and consistent excretion of 3-hydroxyglutaric acid in the urine, a metabolite that has previously been regarded as pathognomonic for GDH deficiency. The underlying genetic defect in these children remains to be identified.

Broad range DGGE using a single, standard gel for all exons rapidly and reliably detected all heterozygous mutations in the present study and should be regarded as the method of choice for GA1 mutation analysis in a routine setting. Homozygous mutations that do not change the melting characteristics are not always recognised using this method, but this is easily circumvented by analysing at least one parent or mixing patient DNA with a normal control DNA sample. However, we did not encounter any mutations in exon 1, which is very GC rich, nor exon 4, and we cannot therefore exclude the possibility that mutations in these exons or other mutations in other exons may be missed by our method. Furthermore, GA1 in some patients may be caused by large deletions or intron mutations that are not susceptible to simple PCR analysis.11

Direct measurement of GDH activity in fibroblasts or lymphocytes remains the diagnostic method of choice as it reflects the pathophysiology in the patient and provides information on residual enzyme activity. However, GDH assays are performed in only a few centres world wide and require the analysis of living cells which may not always be available. In contrast, small amounts of stored DNA or blood are sufficient for molecular analysis, and we have successfully used dried blood spots on old Guthrie cards for mutation detection by DGGE and sequencing. Our method also allows genetic testing in patient families and offers the possibility of reliable carrier analysis in unrelated spouses of GA1 patients. The clinical use of GA1 mutation analysis is restricted by a large number of private mutations of which the impact on GDH activity cannot be predicted. A degree of diagnostic uncertainty remains when novel mutations are found, particularly in patients who do not show the excretion pattern of urine metabolites typical of GA1. In these patients, mutation analysis alone is insufficent for confirmation of diagnosis, and enzyme analysis should also be performed. On the other hand, GDH deficiency in some patients may be caused by defects in the ETF binding site that would be missed by the enzymatic assay but detected by molecular analysis. No such defects have yet been described, but mutation analysis should be performed in all patients with clinical and biochemical diagnosis of GA1 but normal enzyme studies.

Conclusion

The DGGE method presented here may be used for the rapid and cost efficient molecular diagnosis of GA1, particularly when GDH enzyme analysis is not available or feasible. Comprehensive results in the family can be obtained within as little as one week, and the mutation information is particularly useful when prenatal diagnosis is considered for future pregnancies. It remains to be seen whether the wide range of genotypes observed in GA1 can be correlated to differences in residual enzyme activity, clinical features, and outcome.