Tuberous sclerosis (TSC) is an autosomal dominant disorder characterized by tumour-like malformations (hamartomas) of the skin, brain, heart, kidneys and other organs. Half of patients are mentally retarded and most have seizures with one or more of the characteristic dermatological lesions: facial angiofibromas, hypopigmented macules, forehead plaques, shagreen patches and ungual fibromas (1 ). Common hamartomas occurring in TSC include cortical tubers, sub-ependymal nodules, retinal astrocytic hamartomas, renal angiomyolipomas and cardiac rhabdomyomas. Giant cell astrocytomas also occur. The birth incidence is 1 in 10 000 or higher (2 ), with two-thirds of cases being sporadic and representing new dominant mutations.
Tuberous sclerosis shows locus heterogeneity. Approximately equal numbers of families are linked to the TSC1 gene on chromosome 9q34 and the TSC2 gene on chromosome 16p13.3 (3 ). The TSC1 gene encodes a protein (hamartin) which has no homology to proteins of known function (4 ). The TSC2 gene product (tuberin) has homology to the GTPase activating protein rap1GAP (5 ). Tuberous sclerosis hamartomas show loss of heterozygosity at TSC1 or TSC2, supporting the hypothesis that these genes have a role in controlling cellular growth and differentiation (6 -8 ).
Tuberous sclerosis shows wide variation in expression but in segregating families non-penetrance is rare (9 ). Also rare are reports of more than one affected child being born to apparently unaffected parents (2 ,10 -16 ). This could be due to non-penetrance in one of the parents, but the alternative explanation would be germline mosaicism. We now report the results of a molecular genetic analysis which confirms germline mosaicism in a tuberous sclerosis family.
In a family from Northern Ireland three children from a sibship of nine were affected by tuberous sclerosis (Fig. 1 ). The parents were normal with no family history to suggest TSC. None of the affected subjects had offspring; all six of the unaffected siblings had children, 15 in total (11 over 5 years of age), with no evidence of TSC. The affected subjects all fulfilled the diagnostic criteria for tuberous sclerosis proposed by Gomez (18 ). II-3 (seen at 41 years of age), one of dizygotic twins, had suffered from seizures since infancy and had severe mental retardation. Examination showed facial angiofibromas, several shagreen patches on the back and ungual fibromas on most toe nails. II-7 (age 35 years) had facial angiofibromas, a forehead fibrous plaque, several shagreen patches on the back, three gingival fibromas and several dental pits. II-8 (age 31 years) had suffered from seizures since infancy. Examination showed facial angiofibromas, a forehead fibrous plaque, three hypopigmented macules on the legs, several shagreen patches on the back, ungual fibromas on several finger and toe nails, a gingival fibroma and a large mulberry-like retinal astrocytic hamartoma just above the left optic disc. The parents were healthy and unrelated, with no medical history to suggest tuberous sclerosis. Examination of the father (I-1, age 74 years) did not show any dermatological, dental or ophthalmological signs of TSC. The lateral third of the nail was missing on the left fifth toe nail, leaving a fleshy nail bed but no evidence of an ungual fibroma. The mother (I-2) had had one seizure associated with eclampsia in her first pregnancy. Examination at age 69 years did not show any dermatological or dental signs of TSC. Fundal examination showed a small star-shaped pigmented lesion lateral to the right optic disc which was not considered significant. In both parents CT scans of the brain and kidneys were normal. None of the unaffected children had any medical history to suggest TSC. Examination of II-2 (at age 41 years), II-4 (40 years), II-5 (38 years), II-6 (36 years) and II-9 (29 years) did not show any dermatological, dental or ophthalmological signs of TSC. II-4 had five 2-3 mm ill-defined hypopigmented spots on the back of the right leg which were not considered significant. CT scans of the brain were normal in II-2, II-6 and II-9; renal CT scans were normal in II-6 and II-9.
Typing of markers spanning the TSC1 locus at 9q34 showed that two of the affected children (II-7 and II-8) had inherited entirely different maternal and paternal haplotypes (Fig. 1 ). This excluded the TSC1 gene as the cause of the disease in this family. For markers spanning the TSC2 locus at 16p13.3 all affected subjects had the same maternal and paternal haplotypes and so did three of their unaffected siblings (II-2, II-6 and II-9). The father (I-1) was informative for eight microsatellite markers and also the minisatellite markers D16S85 (3'-HVR) and D16S309; there was no incompatibility with the genotypes of his children, providing strong support for paternity.
Southern analysis using TSC2 cDNA probes gave normal results. RT-PCR and direct sequencing of the TSC2 gene identified a 4 bp insertion TACT following nucleotide 2077 in exon 18 present in the affected children. The signal for the mutant sequence was weaker than the normal sequence, suggesting a lower level of expression. PCR amplification of exon 18 from genomic DNA confirmed the presence of a larger fragment in all three affected children which was not present in their five unaffected sibs or the parents (Fig. 2 a). Isolation, amplification and direct sequencing of the normal and abnormal fragments provided independent confirmation of the mutant sequence (Fig. 3 ).
Results of typing an expressed EcoRV polymorphism in exon 40 of the TSC2 gene (18 ) are shown in Figure 1 . All three affected subjects were heterozygous, having inherited allele 1 (restriction site present) from their mother and allele 2 (restriction site absent) from their father. To determine the parental origin of the mutation, PCR of exon 18 was carried out in affected subjects II-7 and II-8. The template for the reaction was a 3 kb PCR product spanning exons 18-40 of the TSC2 gene amplified from mRNA. Both the normal 151 bp and mutant 155 bp fragments were demonstrated, together with heteroduplex bands (Fig. 2 b, lane 3). However, when the template was digested to completion with EcoRV, cleaving fragments derived from the maternal allele which had the polymorphic EcoRV restriction site in exon 40, and the DNA then diluted and reamplified to give a template derived only from the paternal allele, PCR of exon 18 gave only the normal 151 bp fragment (Fig. 2 b, lane 2). This showed that in the affected subjects the TSC2 mutation was in the maternally derived allele.
The 4 bp insertion identified in this family causes a frameshift and premature termination at codon 703. This is in keeping with other reports of mutations in TSC2 resulting in a truncated or absent protein (5 ,19 -21 ). Absence of the 4 bp insertion in lymphocyte DNA from the parents and unaffected children is consistent with germline mosaicism. This is confirmed by the finding that three of the unaffected children (II-2, II-6 and II-9) had identical maternal and paternal chromosome 16 haplotypes to their affected siblings, providing unequivocal evidence of two different cell lines in the gametes. The alternative explanation, that one of the parents and the three unaffected children with the high risk haplotype were all constitutional carriers of the mutation but non-penetrant, is incompatible with repeated failure to demonstrate the mutant allele by PCR analysis of genomic DNA and implausible given that non-penetrance in segregating TSC families is so rare that there is only one reported family where this is a possibility (9 ).
There are eight reports of siblings with tuberous sclerosis born to apparently unaffected parents (2 ,10 -16 ), although in only six of these was exclusion of TSC in the parents supported by brain CT scanning. There is also a report of paternal half-siblings with tuberous sclerosis and polycystic kidney disease whose father was apparently unaffected (22 ). Recurrence of TSC in offspring of unaffected parents could be due to multiple independent new dominant mutations, multiple non-paternity or an autosomal recessive form of the disease, but the most likely explanations are non-penetrance in a parent or germline mosaicism. It has generally been assumed that germline mosaicism accounts for most if not all such cases, because non-penetrance in segregating TSC families is so rare. Our findings provide support for this view by formally demonstrating that germline mosaicism occurs in this condition.
Germline mosaicism was first postulated by Bowen (23 ) to explain the occurrence of the fully penetrant condition achondroplasia in two sisters born to normal parents. This is a rare but important phenomenon (24 ) which occurs in several autosomal dominant and X-linked recessive disorders, including osteogenesis imperfecta congenita (25 ), neurofibromatosis type 1 (26 ) and Duchenne muscular dystrophy (27 ). Depending on where the mutation arises, it is theoretically possible that the mosaicism could be confined to the germline or it could also be present in somatic cells of the parent. In tuberous sclerosis one case of proven somatic mosaicism has been reported (28 ).
We have shown that the germline mosaicism was present in the mother in our family. Male germline mosaicism has been well documented in both autosomal dominant and X-linked disorders. Female germline mosaicism has been reported in several X-linked conditions, particularly in Duchenne muscular dystrophy (29 ), but in autosomal dominant disorders the few reports of female germline mosaicism for identified gene mutations have all been associated with somatic mosaicism. For example, Constantinou et al. (30 ) reported a female with somatic and germline mosaicism for a mutation in the COL1A1 gene and there are well-documented cases of female somatic and germline mosaicism for rearrangements at the D4F104S1 locus associated with facioscapulohumeral muscular dystrophy (31 ). The present family appears to be the first example of female germline mosaicism for a characterized autosomal dominant gene mutation without evidence of somatic mosaicism. PCR analysis of genomic DNA directly extracted from peripheral blood leukocytes showed no evidence of the mutation in the mother. However, samples of other tissues to allow more comprehensive exclusion of somatic mosaicism were not available.
In conclusion, we have described the first case of germline mosaicism in tuberous sclerosis which has been proven by molecular genetic analysis. This has important implications for genetic counselling. For apparently isolated cases of tuberous sclerosis where the parents have been fully investigated, including brain CT or MRI scanning, and appear unaffected, they are still at risk of another affected child. On the limited data available the recurrence risk is estimated at ~2%. The family we have reported is also of more general biological interest, showing that germline mosaicism for an autosomal dominant gene mutation can occur in females apparently in the absence of somatic mosaicism.
All family members with the exception of II-1 were examined by J.R.W.Y. and D.W.W., including inspection of the skin with a Woods lamp, direct ophthalmoscopy with dilated pupils and examination of the teeth after application of liquid dental disclosing dye to highlight enamel pits.
Leukocyte DNA extraction was carried out using standard methods. The family was typed for the markers ASS, D9S149, D9S1198, D9S150 and D9S114 spanning the TSC1 locus on 9q34 and D16S283, D16S665, KG8, D16S309, D16S525, D16S85 (3'-HVR) and HBAP1 spanning the TSC2 locus on 16p13.3. The family was also analysed for an expressed EcoRV polymorphism in exon 40 of the TSC2 gene as previously published (18 ).
Affected family members were screened for large deletions and rearrangements by Southern analysis using TSC2 cDNA probes. This was followed by screening for small deletions by RT-PCR and direct sequencing. Total RNA was extracted from lymphoblastoid cell lines using TRIzoltm (Life Technologies) according to the manufacturer's instructions. cDNA was synthesized using the Superscripttm Preamplification System (Life Technologies). All cDNA was amplified using the provided oligo(dT) primers. To make templates for sequencing, three PCR products were generated, designed to overlap by at least 80 nt, spanning the entire coding region of the TSC2 gene. Primer sequences and annealing temperatures have been published elsewhere (21 ).
To amplify exon 18 from genomic DNA, PCR primers were designed from the published TSC2 cDNA sequence (5 ) and information about intron-exon boundaries (32 ). The forward and reverse primers were 5'-GGAGCCAGAGAGAGGCTCTGA-3' and 5'-CTGCTTCAAGCACTGCAGCGG-3' respectively. PCR reactions were carried out for 35 cycles in a total reaction volume of 25 µl (95°C 30 s, 64°C 30 s, 72°C 30 s). After PCR amplification the products were analysed on a 10% PAGE gel and visualized by silver staining. In order to sequence the normal and mutant alleles individually, they were separated on a 10% PAGE gel, excised and purified. The purified products were used for a second round of the same PCR. These products were used for direct automated sequencing as described above.
Long range PCR was used to amplify TSC2 mRNA from affected subjects II-7 and II-8. The 3 kb product spanned exons 18-40 with the polymorphic EcoRV restriction site 180 bp from the 5'-end. The PCR product was digested to completion with EcoRV and purified. A dilution was then reamplified using the same primers. An aliquot of the resulting product was typed for the EcoRV polymorphism to confirm that only the allele lacking the EcoRV restriction site (allele 2 derived from the father) had been amplified. The PCR product was then analysed for the presence or absence of the 4 bp insertional mutation by PCR analysis of exon 18 as described above.
We are grateful to the family for co-operating with this study. We thank Dr Patrick Morrison for help with making the arrangements for us to visit the family. This work was supported by grants from the Westminster Medical School Research Trust (Royds Fund) (I.v.B) and the Tuberous Sclerosis Association of Great Britain (T.S.).
*To whom correspondence should be addressed at: Department of Medical Genetics, Box 134, Addenbrooke's Hospital NHS Trust, Cambridge CB2 2QQ, UK. Tel: +44 1223 216446; Fax: +44 1223 217054; Email: jyates@hgmp.mrc.ac.uk
Human Molecular Genetics
Pages
Introduction
Results
Clinical data
Genetic marker analysis
Mutation analysis
Parental origin of the mutation
Discussion
Materials And Methods
Patients
Genetic marker analysis
Mutation analysis
Parental origin of the mutation
Acknowledgements
References
REFERENCES
Present addresses: +Hexagen, 214 Cambridge Science Park, Milton Road, Cambridge CB4 4WA, UK,
§The Kennedy Galton Centre for Medical and Community Genetics, Northwick Park and St Mark's NHS Trust, Harrow, Middlesex HA1 3UJ, UK and
¶National Centre for Medical Genetics, Department of Paediatrics, University College Dublin, Our Lady's Hospital for Sick Children, Crumlin, Dublin 12, Ireland
This page is maintained by OUP admin. Last updated Sat Oct 18 13:29:50 BST 1997. Part of the OUP Journals World Wide Web service.
Copyright
Oxford University Press, 1997