Article Text

Download PDFPDF

A constitutional homozygous mutation in theRB1 gene in a patient with unilateral retinoblastoma

Statistics from

Editor—Retinoblastoma, a childhood tumour of the eye, is caused by inactivation in the developing human retina of both alleles of the tumour suppressor gene RB1. The product of the human RB1 gene (p110RB) is a nuclear phosphoprotein composed of 928 amino acids, which regulates the progression through the G1 phase of the cell cycle by interacting with transcription factors required for the expression of genes involved in cellular proliferation and differentiation.

In the non-hereditary form of the disease (∼60% of tumours), both mutations arise in retinal cells. Because in non-hereditary RB patients both RB1 mutations must occur in the same retinal cell, they are usually unilateral and unifocal. However, based on retrospective surveys of the offspring of patients with unilateral isolated retinoblastoma, Vogel1 estimated that in 10-12% of these patients the tumour is caused by germline cell mutations.

In the hereditary form of the disease (∼40% of RB patients), the initial mutation in one allele of the RB1gene is present in germline cells and leads to a predisposition to retinoblastoma. Since mutations in the second allele can occur independently in several retinal cells carrying this predisposing mutation, all patients with bilateral or unilateral multifocal RB are classified as having hereditary retinoblastoma. Most hereditary cases must result from “de novo” germline mutations, because only 10-15% of the hereditary cases have a previous family history of the disease. In familial cases, the predisposition to RB is transmitted as an autosomal dominant trait with 90% penetrance. Thus, in some familial cases, unaffected or only unilaterally affected subjects can be identified who can transmit the mutant gene.1 The “two hit” hypothesis predicts the existence of these cases that form part of a Poisson distribution in which, by chance, the second random mutation does not occur.2 However, the distribution of cases of incomplete penetrance is not entirely random, and families have been reported in which the majority of the carriers have either unilateral tumours, regressed tumours, or no evidence of malignant disease.3-6 Some hypotheses have been set forth to explain this phenomenon: the existence of “delayed mutations,7 the “host resistance” model,8 9 and the existence of lethal alleles at the cellular level.10 11 Cloning and characterisation12 13 of theRB1 gene made it possible to determine the nature of the mutations, and it has been shown that the mutations found in families with low penetrance retinoblastoma, rather than being stop codons or mutations which are presumed to abolish p110RBprotein activity completely, as described in severe cases of retinoblastoma, are either missense mutations,5 in frame deletions,14 or mutations affecting the promoter region of the gene,6 15 which generate a partially defective protein (low penetrance mutations).16 17 In addition to the low penetrance mutations, other causes of incomplete penetrance and reduced expressivity have been described in retinoblastoma: the existence of mosaicism18 and the aggregation of sporadic genetic events in the same family (“pseudo low penetrance”).14 19 20

We present here a family with three carriers of a mutation in the donor splice site of intron 5 (G→A). This change causes a protein lacking only exon 5 because it does not disturb the frameshift. Only one of the carriers is affected by unilateral retinoblastoma. Furthermore, the affected child has mosaicism comprising homozygosity and heterozygosity for the mutation. The characteristics of the mutation, its location, and the existence of unaffected carriers in the family allow us to hypothesise that this is a low penetrance mutation.

The patient in this study was admitted to hospital at 11 months of age as he had leucocoria in his right eye. His mother noticed it first when he was 2-3 months old and he underwent ophthalmological study eight months later. An analysis of the fundus of the right eye showed that the optic nerve could not be seen because detachment of the retina had taken over three quarters of the vitreous cavity. Ophthalmoscopic examination showed a typical exophytic retinoblastoma. The eye globe was 1.8 cm in diameter (maximum) and inside was observed a retinoblastoma which did not infiltrate the choroid, optic nerve, or ciliary body. The child is now 32 months old and the healthy eye has been checked every six months without any change. Ophthalmological study of the patient's father showed no tumour and the fundus of both eyes was normal.

Samples of peripheral blood and tumour tissue were obtained from the child when he was 12 months old. He was diagnosed at 11 months as having a unilateral sporadic retinoblastoma and his affected eye was enucleated. Blood samples were also obtained from his unaffected parents and grandparents. Genomic DNA from leucocytes and fresh tumour tissue was isolated by using standard phenol/chloroform procedures. Total RNA from these tissues was obtained by means of the RNeasy kit (Quiagen) following the recommendations of the manufacturers.

In order to carry out the haplotype analysis, genomic DNA from white cells and tumour tissue was subjected to Southern blot analysis or PCR amplification or both to genotype RFLP and microsatellite markers within and outside the RB1 gene (table 1). All the intragenic polymorphic markers were detected as indicated in the references. The exception was the Rbi2 and Rbi4 markers, which were amplified by a two step PCR (10 cycles at 94°C for one minute and 65/55°C for one minute with a decrease of 0.5°C/cycle, followed by 20 cycles at 94°C for one minute and 60/50°C for one minute with a final extension at 65°C for five minutes). The extragenic markers were amplified for 35 cycles at 94°C for 40 seconds and 55°C for 30 seconds and a final extension at 72°C for two minutes.

Table 1

Intragenic and extragenic polymorphic markers used for the haplotype analysis

Densitometric analysis was performed using theBamHI polymorphism located in intron 1 of the RB1 gene. This polymorphic marker is detected by the p123M1.8 probe, which identifies both the 5′ end of theRB1 gene and a band corresponding to a fragment of the 28S rDNA gene.21 We used this constant band as an internal control. Quantification of the genetic doses of theBamHI RFLP alleles was performed using computer software (Intelligent QuantifierTM, Bio Image®) that measured the intensities of bands from digitised images (Gelstation, TDI).

In order to perform screening for small mutations and sequencing, the promoter region and 27 exons nearest the intronic regions of theRB1 gene were PCR amplified using the primers described by Hogg et al 22 and Shimizu et al.23 PCR products were digested with the appropriate restriction enzymes to produce DNA fragments of 250 bp or less. Mutation screening was performed by SSCP analysis and samples showing altered electrophoretic behaviour were subjected to direct sequencing.

RNA from peripheral blood and tumour tissue was reverse transcribed using the RT-PCR kit (Stratagene) and following the recommendations of the manufacturers. The resulting first strand cDNA was PCR amplified using primers which were designed to amplify six overlapping fragments of the coding sequence from the RB1gene.24 The distribution of exons in these fragments was: fragment 1, exons 2-6; fragment 2, exons 6-10; fragment 3, exons 10-16; fragment 4, exons 16-19; fragment 5, exons 19-23; and fragment 6, exons 23-27. Amplification conditions consisted of 35 cycles at 94°C for one minute, 55°C for one minute, and 72°C for one minute, and a final extension at 72°C for 10 minutes. Individual PCR products were subjected to fragment length analysis, SSCP analysis, and direct sequencing. For the SSCP, digestion of the PCR products 1 to 6 was carried out before analysis with enzymesRsaI, ApaII,DraI, DdeI,NdeI, and ApoI, respectively.

Molecular analysis of the RB1 gene in the family studied showed a decrease in the dose of the maternal alleles in intragenic polymorphic markers at a constitutional level in the RB-327 patient (fig 1). The densitometric analysis of theBamHI marker (fig 2A) showed that, while the internal control band of a normal heterozygote and that of the affected child had similar intensities, the peak corresponding to allele 1 in the affected child (paternally derived allele) had a higher intensity than the heterozygous control in DNA isolated from leucocytes. This increase in the paternal allele was associated with a decrease in the dose of the maternally derived allele (allele 2) with regard to the heterozygous control sample (fig 2B). Analysis of the patient's tumour DNA showed an undetectable maternal allele (LOH), while the paternal allele was similar to that found in a homozygous control (fig 2C). The same loss of the maternal allele was observed for other polymorphic markers on chromosome 13 flanking the RB1gene, except for the most centromeric one (D13S787), in which both leucocytes and tumour DNA showed the maternal and paternal alleles at a similar level (fig3).

Figure 1

(A) Pedigree of the RB-55 family. The identification number appears below the symbol for each subject. (B) Southern blot analysis of the VNTR marker in intron 17 of the RB1 gene (detected with probe p68RS2.0) showing that the affected child has a less intense maternal allele (arrow).

Figure 2

(A) Results of the RFLP marker BamHI (intron 1 of the RB1 gene) analysis in DNA from peripheral blood of the family and the retinoblastoma tumour of the child (RB-327T). This analysis was performed by means of the intragenic p123M1.8 probe. Individual alleles are either a fragment of 4.5 kb or a pair of fragments of 2.3 and 2.2 kb. A constant band belonging to the rRNA 28S gene was also detected. This band is used as a reference control of the amount of DNA in each row. A heterozygous control (Ht.C) and a homozygous control (Hm.C) have been included. (B) Comparison of the intensities of the control band and alleles of the BamHI marker in the child's blood (black line) and the heterozygous control (grey line). (C) Comparison of the intensities of the control band and alleles of the BamHI marker in the tumour (black line) and the homozygous control (grey line)

Figure 3

Informative extragenic markers studied in this family. Asterisks indicate the maternal allele in the patient. M: 1 kb ladder. The markers are arranged according to their location in 13q and the position of the RB1 gene is indicated.

On the other hand, gel electrophoresis of the RT-PCR product showed a band of normal size and a smaller band when we used the primers to amplify fragment 1 of cDNA from the child and his father's leucocytes (fig 4). The same fragment obtained from tumour mRNA showed only the smaller band. In the maternal sample, a normal length band was detected. Sequence analysis of that anomalous band showed the loss of exon 5 in fragment 1 (fig 4), but this does not alter the reading frame. SSCP analysis of leucocyte DNA detected an abnormal electrophoretic pattern in the exon 5 PCR product in the proband and his unaffected father and grandmother, as well as in tumour DNA. Direct sequencing of this fragment showed a nucleotide change (G→A) at the splice donor site in intron 5 (position 44707 of the sequence shown in GenBank under Accession No L11910) (fig 5). Because the tumour cells and some constitutional cells of the affected child are homozygous for the paternal allele, they carry the mutation homozygously.

Figure 4

(Left) Results of the sequence analysis of fragment 1 amplified from cDNA obtained from a normal control (top) and from the tumour sample (bottom). The vertical broken line indicates the limits of the exons. (Top right) Agarose gel electrophoresis of fragment 1 (which spans exons 2-6 of the RB1 gene) amplified from the cDNA obtained from the lymphocytes of the affected child (RB-327), his parents (RB-330 and RB-331), and a normal control (C+), and from the tumour tissue (RB-327T). Lanes RB-327 and RB-331 show a band of normal size and a smaller anomalous band. In the row belonging to the tumour sample only the anomalous band was observed. (Bottom right) Diagram showing the normal spliced sequence of cDNA and that corresponding to the removal of exon 5 observed in the cDNA from some members of the family and the tumour.

Figure 5

(Left) SSCP analysis of exon 5 in genomic DNA. Arrows indicate the anomalous electrophoretic bands in the samples RB-329, RB-331, RB-327, and RB-327T. (Right) Sequence analysis of exon 5 in tumour tissue. Arrows indicate the position of the change (G→A) detected in the donor splice site in intron 5. Vertical lines in the sequence indicate the limits of exon 5.

To date, many types of oncogenic mutations have been described as causing retinoblastoma (a summary can be seen in htm). These mutations are located along the whole RB1 gene, and the majority of them cause highly penetrant retinoblastoma. We describe a family in which three members are carriers of a point mutation affecting the correct splicing of exon 5 in the mRNA. The fact that the loss of exon 5 does not alter the frameshift and two of the carriers of the mutation are unaffected lends support to the idea that this is not an oncogenic mutation. However, there are also arguments against this assumption. First of all, the screening by SSCP analysis of all other exons and the promoter region of the RB1gene did not show any other mutation in this family, although in all tumours studied by other authors, at least one of the mutations in theRB1 gene has been found. Secondly, some studies of the N-terminal region of the protein encoded by theRB1 gene have shown that deletions in this region disturb its capacity to polymerise itself in order to form compartments to arrest other molecules.25-27 Furthermore, some reports have shown binding of this region of p110RBwith other nuclear proteins like laminin A and C and p84,28-30 representing a potential site of p110RB interaction in the nuclear matrix. The nuclear matrix has been implicated in most metabolic activities occurring in the nucleus, including replication, transcription, and RNA splicing and transport.31-33 p110RB may facilitate the binding of growth promoting factors at subnuclear regions actively involved in RNA metabolism.30 Finally, the third reason we think that the mutation is not a polymorphic variation is that comparison of the amino acid sequences encoded by exon 5 in some species has shown a high degree of conservation, especially in higher vertebrates, and this indicates the importance of this exon.

Although we have only analysed a few subjects in three generations, some of ours findings point to a low penetrance mutation: (1) the fact that only one of the three carriers of the mutation is affected by unilateral retinoblastoma; (2) another mutation (deletion of exon 4) probably with similar consequences in the p110RB protein has been described in a family with low penetrance retinoblastoma,14 where the protein p110RBΔ4was defective for E2F binding but able to activate transcription and promote differentiation34; and (3) the fact that the mutation was found homozygously at a constitutional level. If it were a highly penetrant mutation, it would probably be lethal. Studies done on mice showed that embryos lacking any functional p110RBprotein die in utero by day 14.5 of gestation. p110RB is also required in the terminal differentiation of some tissues.35-38

The results obtained in blood and tumour from the patient with the different markers used (the most centromeric marker showed the maternal and paternal alleles at a similar level, while the rest of the markers showed a loss of the maternal allele) and densitometric analysis (the increase in the paternal allele dose is associated with a decrease in the dose of the maternally derived allele) suggest that the affected child was a mosaic composed of at least of two constitutional cells lines, one of them with a paternal and a maternal allele, and the other with a duplication of the paternal allele.

The mosaicism detected in blood from the patient could have arisen as a consequence of a mitotic recombination event during the earlier stages of embryonic development between markers D13S787 and D13S325 (proximal to the RB1 gene) (fig 3). The fact that the tumour cells are carriers of a LOH with a double dose of the paternal allele for the same markers in which a reduction of the maternal allele has been shown in blood cells indicates that the tumour arose from a cell with the mutation in double dose. This is remarkable given that, to our knowledge, this is the first time that a low penetrant mutation has been described in homozygosity, both at a tumour and constitutional level. Sakai et al 6 hypothesised that a cell homozygous for a low penetrance mutation would not evolve into a tumour because a minimum threshold of p110RB protein activity is retained,5 16 but in fact only a few tumours in low penetrance retinoblastoma patients have been analysed so far.14 39

There are two possible explanations for the development of the tumour in our patient, in spite of his double dose of a low penetrance mutation. One is the role of p110RB in an unknown mechanism in retinal tissue not screened in the expression studies of mutations with low penetrance. The other explanation, more probable in our opinion, is that a mutation in a gene other than theRB1 gene or an undiscovered third mutation in the RB1 gene is necessary for tumour development in this family (three hit model).

To date, only one case of paternal disomy 13 has been described.40 A normal phenotype was observed in that case and an unlikely imprinting effect was postulated for this type of disomy.41 Similarly, the recent clinical assessment of the patient reported here did not show any manifestation other than the retinoblastoma. The serotonin receptor gene (HTR2), closely linked to theRB1 gene, has been reported.42 43 and studies with retinoblastoma tumours have shown only expression of the maternal allele of this gene.44 Serotonin (5-hydroxytryptamine (5-HT)) is a neurotransmitter that mediates a diverse array of physiological responses by interacting with multiple serotonin receptor subtypes. Two of them, 5HT2 and 5H1c, modulate similar intranuclear signalling pathways but exhibit different patterns of expression in the brain. The affected child in the family reported here is a mosaic carrier of paternal disomy of almost the whole of q13, so he probably lacksHTR2 gene expression in a large proportion of his cells. Nevertheless, the patient is too young to assess the putative neurological consequences of the pattern of expression of theHTR2 gene.

All the unusual findings described in our family, that is, mosaicism and variable expressivity and transmission, have important implications for the DNA based estimates of the prognosis of the disease and accurate genetic counselling.


The first two authors contributed equally to this work. We should like to thank the families for their generous cooperation, Dr T P Dryja for providing the genomic probe used in this study, and Drs Serra, Harto, and Castell (Paediatric Ophthalmology and Paediatric Oncology services) for their clinical assistance. This work was supported by the CICYT (SAF 92/0206). F Sánchez is the recipient of a fellowship from the Consellería de Educación y Ciencia de la Comunidad Valenciana.


View Abstract

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.