Editor—The Li-Fraumeni syndrome (LFS) represents one of the most devastating genetic predispositions to cancers. This rare syndrome, affecting children and young adults, is characterised by a wide spectrum of early onset malignancies including bone and soft tissue sarcomas, brain tumours, adrenocortical tumours, and premenopausal breast cancers.1 LFS was initially defined using stringent criteria2: (1) a proband with a sarcoma diagnosed before the age of 45, (2) a first degree relative with cancer before the age of 45, and (3) another first or second degree relative with either a sarcoma diagnosed at any age or any cancer diagnosed under the age of 45. Subsequently, Birch et al 3 defined Li-Fraumeni-like (LFL) syndrome as a proband with any childhood tumour or sarcoma, brain tumour, or adrenocortical tumour under 45 years, plus a first or second degree relative with a typical LFS tumour at any age and another first or second degree relative with any cancer under the age of 60. Eeles4 proposed more relaxed criteria for LFL: a clustering of two typical LFS tumours in subjects who are first or second degree relatives at any age. Since the original reports of germline mutations of the tumour suppressor geneTP53 in LFS,5 6 numerous studies have shown that germline TP53mutations can be detected in approximately 70% of LFS families and 20% of LFL families,1 suggesting the possible involvement of other genes in LFS. This hypothesis was recently confirmed by the detection, in one LFS family and one family suggestive of LFS, of germline mutations of hCHK2, the human homologue of the Saccharomyces cerevisiae RAD53 gene, located on chromosome 22q12.7 8 hCHK2 encodes a kinase, which is able to phosphorylate, in response to DNA damage, the Cdc25C phosphatase involved in the G2 checkpoint7 and TP53.9-11We considered that the TP53 homologue,TP63/p51/KET 12-14 located on 3q27-29, was also a candidate gene for the LFS not associated with the germline TP53 mutation, since the isotype p63γ transactivates reporter constructs containing theTP53 binding sites present withinp21/WAF1/CIP1/CDKN1A,BAX,MDM2,15 andPIG3 (G Bougeard, T Frebourg, unpublished data) genes.

We therefore analysed in this study theTP53, hCHK2, andTP63 genes in 17 LFS/LFL families referred to our laboratory. These 17 families (table 1, figs 1 and 2) included four families fulfilling the complete criteria for LFS, five the criteria for LFL, as defined by Birch et al 3 and were therefore designated LFLb, and eight the criteria for LFL proposed by Eeles4 and were therefore designated LFLe. For 15 index cases, we performed theTP53 functional assay developed in yeast on cDNA derived from lymphocytes.16 17 Briefly, the reporter yeast strain yIG397-RGC was co-transformed with PCR amplifiedTP53 cDNA (between codons 53 and 364) and the gapped expression vector pSS16 linearised between codons 67 and 346, and cDNA was cloned in vivo by homologous recombination. The activation by wild type TP53 of the reporter system, containing the ADE2 open reading frame and theTP53 binding site RGC, changes the colour of the yeast colonies (red→white) and samples containing wild typeTP53 commonly give a background of 5-8% red colonies owing to PCR induced errors and the presence of an alternatively spliced TP53mRNA.16 17 In 10 families, the percentage of red colonies (above 10%) suggested the presence of a heterozygous mutation, which was confirmed by sequence analysis of cDNA and/or genomic DNA (table 1, fig 1). Sequencing of exons 2 to 11 confirmed the absence of mutation in the five families with a normal TP53functional assay and, in one of the two families for which no RNA was available, allowed the detection of an additional mutation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Partial pedigrees of 11 LFS/LFL families with a germline TP53 mutation. Filled symbols=affected subjects, open symbols=asymptomatic subjects. For each affected subject, the tumour (or site of the tumour) and age of diagnosis are indicated. The age of unaffected subjects is indicated in brackets. Adeno: adenocarcinoma of unknown origin; adreno, adrenocortical carcinoma; astro, astrocytoma; epen, ependymoma; GI, gastrointestinal tumour; histio, histiofibrosarcoma; lipo, liposarcoma; med-epith, medullo-epithelioma; medullo, medulloblastoma; oligo, oligodendroglioma; osteo, osteosarcoma. Germline TP53 mutations are shown in table 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Partial pedigrees of six LFL families with no detectable germline TP53 mutation.

In the six families without detectable TP53germline mutation (fig 2), we then analysed theTP63 gene. To screen for inactivatingTP63 mutations, we performed a functional assay in yeast. Transformation of the pCI51 plasmid18containing the wild type coding region corresponding to the isoform γ of TP63 into the yIG397-RGC yeast reporter strain resulted in white colonies. Transformation of two mutant pCI51 plasmids (with the mutation Leu264Ser or Cys269Ser generated by PCR induced mutagenesis), used as controls, resulted only in red colonies. In four index cases, for whom mRNA was available, the TP63 cDNA was PCR amplified from lymphocytes between codons 18 and 434, cloned by homologous recombination into the gapped expression vector pCI51 linearised between codons 30 and 420, and transformed into yIG397-RGC; the percentage of red colonies (table 1) suggested the absence of heterozygous inactivating mutations. We also sequenced in the six families without detectable TP53 mutation exons 2 to 15 of the TP63 gene from genomic DNA, using primers described by Hagiwara et al,19 and detected no nucleotide change. We then analysed the open reading frame of hCHK2 by RT-PCR in the four index cases for whom mRNA was available. As recently highlighted by Sodha et al,8exons 10 to 14 of hCHK2 have homologous fragments on numerous chromosomes, which limits analysis from genomic DNA, and the mutation screening should therefore be performed by RT-PCR. The entire coding region of hCHK2(1662 bp) was PCR amplified from cDNA derived from lymphocytes using primers (5′-TGT CTC GGG AGT CGG ATG TTG AGG CTC AGC-3′) and (5′-GGA CAT TTC TTT CGT GTT CAA ACC ACG GAG-3′), and the PCR products were then submitted to a second stage PCR amplification generating three overlapping fragments which were sequenced. In all subjects, RT-PCR showed an in frame deletion of 87 bp, which was also observed in controls. Alignment, using the BLAST program (National Center for Biotechnology Information), of the hCHK2cDNA (accession number AF086904) to the genomic clones RP11-44G7 (accession number AL117330) and RP11-436C9 (accession number AL121825) derived from chromosome 22 showed that the deleted fragment corresponds to an exon (probably exon 9), indicating the presence of an alternative splicing of hCHK2 in lymphocytes. Sequence analysis detected no germline hCHK2 mutation in the four probands.

We therefore identified in this study 11 germlineTP53 mutations, including six previously unreported germline mutations (table 1), in 4/4 LFS, 4/5LFLb, and 3/8 LFLe families (fig 1), confirming that the use of relaxed criteria decreases the probability of identifying a germline mutation. This study shows that the functional assay is an efficient method to detect germline TP53 mutations since, in this series, it detected all the germline mutations. The recent identification in LFS families of germline mutations ofhCHK2,7 encoding a kinase whose substrates include Cdc25C20 and TP539-11strongly suggests that the critical defect in LFS is the constitutional alteration of the G1 and/or G2 checkpoints in response to DNA damage. In four LFL families, with no detectable germlineTP53 mutation, we did not detect any germline mutation of hCHK2. In this study, we also explored the involvement of TP63, one of the TP53 homologues identified over the last two years. Somatic TP63 mutations appear very rare in tumours.18 19 Nevertheless, when we performed a functional assay in yeast on aTP63 cDNA submitted to PCR induced mutagenesis, as previously described,21 we were able to estimate the number of sites in which mutations could be detected as 528, indicating that TP63 exhibited in vitro the same intrinsic sensitivity to mutations asTP53. Our study provides no evidence for the involvement of TP63 in LFS. In contrast, germline TP63 mutations were recently identified in the EEC (ectrodactyly, ectodermal dysplasia, and cleft lip) syndrome.22 It could be argued that, in this study, the six families with no detectable mutation correspond either to other inherited forms of cancers or to the aggregation of sporadic cancers. Nevertheless, the association of early breast cancer with sarcoma or a CNS tumour in first degree relatives, observed in families F13, F14, and F15 (fig 2), is strongly suggestive of LFS. Therefore, our results indicate the involvement of other genes in LFS. If the medical benefit of presymptomatic testing in these families is not obvious considering the wide spectrum of tumours, in contrast identification of the germline alteration in affected subjects confirms on a molecular basis the diagnosis of LFS, which may have important clinical implications. These patients, who have a high risk of developing multiple primary cancers,23 may benefit from a regular clinical review. Furthermore, several studies have reported, in mutation carriers, the development of second tumours in the radiotherapy fields, which raises the question of the use of ionising radiation in these patients.23 Therefore, the complete characterisation of the molecular basis of LFS will be important for the correct clinical management of these families.