Mutation of the RB1 Gene Caused Unilateral Retinoblastoma in Early Age

Mutation of the RB1 Gene Caused Unilateral Retinoblastoma in Early Age Judit Damjanovich, Róza Ádány, András Berta, Zoltán Beck, and Margit Balázs

ABSTRACT: Fluorescence in situ hybridization (FISH) was applied for the detection of the retinoblastoma tumor-suppressor gene deletion on retinoblastoma tumor cells obtained from the unilateral tumor of a 3-month-old boy. Both retinoblastoma tumor cells and peripheral lymphocytes of the patient showed one hybridization signal per cell at the retinoblastoma-1 locus, indicating that one copy of the gene was deleted. Peripheral blood lymphocytes obtained from the patient’s parents had two copies per cell for the gene. Retinoblastoma nuclear phosphoprotein expression could not be detected in the tumor tissue. No copy number alterations were detected with ten different centromeric DNA probes in the tumor cells. The deletion at the RB1 locus detected by FISH suggested that this gene alteration was heritable. The parental peripheral blood lymphocytes did not show the loss of the gene; thus the first deletion may have taken place in either of the parental germ cells. The second somatic mutation of the RB1 gene was probably under the detection limit of FISH. The second allelic alterations were detected by using the polymerase chain reaction for all exons of the retinoblastoma gene. © 2000 Elsevier Science Inc. All rights reserved. INTRODUCTION Retinoblastoma is an embryonic tumor originating from precursors of photoreceptor cells of the neural retina and can appear in sporadic or familial forms [1]. The hereditary form develops at an earlier age than the sporadic one and usually appears as bilateral or multicentric tumors or both. Approximately 80% of the unilateral cases are a result of a somatic, nonhereditary mutation [2, 3]. Knudson [4] was the first to observe that a deletion of the long arm of chromosome 13 was associated with the disease and that two mutations had to be present for the manifestation of the tumor. Detailed molecular genetic studies have revealed that almost all retinoblastomas show homozygous loss of function of the retinoblastoma susceptibility gene (RB1) at the 13q14 locus [2]. The RB1 gene is a well-characterized tumor-suppressor gene that encodes a cell cycle nuclear phosphoprotein [5]. The gene is composed of 27 exons dispersed over about 200 kb of genomic DNA [6].

From the Department of Ophthalmology (J. D., A. B.), the Department of Hygiene and Epidemiology (R. Á., M. B.), and the Department of Microbiology (Z. B.), Medical University School of Debrecen, Debrecen, Hungary. Address reprint requests to: Judit Damjanovich, Department of Ophthalmology, University Medical School of Debrecen, Debrecen, Nagyerdei krt. 98, H-4012 Hungary. Received November 5, 1998; accepted August 24, 1999. Cancer Genet Cytogenet 119:1–7 (2000)  2000 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010

Various techniques were proposed for the screening of RB1 gene alterations, but no hot spots for mutations have been found until now [7]. Standard karyotyping, which can be applied only to mitotic cells, enables the identification of deletion but has a limited resolution. Fluorescence in situ hybridization (FISH) is a well-established technique for identifying gene deletions in interphase nuclei and metaphase chromosomes [8, 9]. The deletion of the RB1 gene was shown to often involve a large region of the 13q14 locus [9, 10]. Therefore, FISH is a useful and reliable method for screening constitutive deletions in retinoblastoma patients and their relatives by using the 13q14 RB1 gene-specific probe. FISH has been used to define RB1 deletions in different tumor types [9, 11, 12] but was reported to be applicable in retinoblastoma tumors in only a few cases [9]. In this study, FISH was used to characterize the RB1 gene alteration in retinoblastoma tumor cells obtained from a 3-month-old boy. Interphase cytogenetic data of the tumor cells were compared with hybridization data on lymphocytes obtained from the patient and his parents’ peripheral blood by using RB1- and centromere-specific DNA probes. According to the “two hit” theory, inactivation of both functional copies of the 13q14-associated RB1 gene must be demonstrated. The FISH technique has a limit in detecting small alterations. To avoid sequencing of the complete coding sequence, several prescreening methods have been developed and applied to the RB1 gene. Among them, single-strand conformation polymor-

0165-4608/00/$–see front matter PII S0165-4608(99)00198-3

2 phism (SSCP) based on a polymerase chain reaction (PCR) appears to be highly sensitive and suitable for detecting small changes in the genome [13–16]. For this reason, an exon-by-exon PCR-completed SSCP was used to analyze the retinoblastoma gene.

MATERIALS AND METHODS Tissue Samples and Cell Preparation This study was carried out in full accord with the principles of the Declaration of Helsinki and was approved by the Human Experimentation Committee of our university. The patient was a 3-month-old boy, with no family history of retinoblastoma. Only his left eye was affected by the tumor, invading more than two-thirds of the eyeball; his right eye was healthy. The tumor caused leukokoria, which was first recognized by the patient’s mother. The tumor appeared to be unifocal, and there was no sign of any tumor symptoms in the other eye during the 41-month follow-up period. The patient was otherwise healthy and free of any deformations in the midface [17]. It was the early onset that suggested a hereditary origin of the tumor. After surgical removal, the enucleated globe was cut into two parts. A small piece removed from the tumor tissue was used to obtain a single-cell suspension for FISH. Cells were washed in phosphate-buffered saline (PBS), hypotonized in 75 mM KCl (15 min at 37⬚C), and fixed in Carnoy’s solution (3⬊1 methanol⬊acetic acid). The remaining part of the tumor tissue was fixed in 4% paraformaldehyde for 6 hours and embedded in paraffin. Lymphocytes from the patient’s and his parents’ blood were prepared by using standard protocols [9, 11]. Control lymphocytes were obtained from a healthy volunteer [11]. Neuron-Specific Enolase and RB Protein Detection The tumor tissue was examined for neuron-specific enolase (NSE) expression to make sure that the cells used for FISH were tumor cells [18]. Five-micrometer-thick sections were cut and mounted on gelatin-coated slides. Then 5% normal human serum in PBS was used for 15 minutes to block the nonspecific antibody binding. The tissue sections were incubated with mouse-anti-NSE monoclonal antibody (1⬊50; DAKO, USA) for 2 hours at room temperature (RT). The antigen–antibody reaction was visualized with biotinylated anti-mouse immunoglobulin G (IgG; 1⬊200; 30 min, Vector Laboratories, Burlingame CA, USA) and a 1⬊40 dilution of Texas Red-Streptavidin (30 min, Vector Laboratories). The sections were counterstained with 4⬘,6-diamino-2-phenylindole hydrochloride (DAPI, Molecular Probes, Eugene, OR, USA) at 0.02 ␮g/mL to show nuclei. The tumor tissue was also examined for RB gene expression [19]. An immune reaction for the RB protein was carried out on formalin-fixed, paraffin-embedded tissue sections. Human embryonic tissue samples (skin and cartilage) were used as controls. The slides were incubated with an optimal dilution of monoclonal antibody (1⬊10 dilution, 1 hr, Novocastra NCL-RB). The antigen–antibody reaction was detected using a Vectastain ABC kit according to the supplier’s protocol. (Vector Laboratories).

J. Damjanovich et al. Chromosome-Specific DNA Probes Centromeric DNA probes (University of California, San Francisco, CA, USA) were used for chromosome 1 (pUC1.77), chromosome 7 (p7␣tet), chromosome 9 (pUH98), chromosome 11 (pLC11A), chromosome 15 (pD1521), and chromosome 17 (p17H8). The probes were biotinylated by nick translation according to the protocol of the supplier (GIBCO-BRL, Gaithersburg, MD, USA). Biotinylated chromosome-specific DNA probes for chromosomes 13/21, 20, X, and Y were obtained from Oncor. RB1 gene-specific probes were purchased from two different companies. A biotinylated locus-specific probe for the 13q14 was from ONCOR; the Spectrum Orange-labeled 13q14 probe and the LSI 13 Spectrum Green/LSI 21 Spectrum Orange probes were from Vysis (Vysis Inc.). Fluorescence In Situ Hybridization Fluorescence in situ hybridization was carried out on cell preparations as previously described [12]. Briefly, Carnoy’s fixed nuclei were denatured in 70% formamide/2 ⫻ SSC (0.3 M NaCl, 0.03 M Na citrate), pH 7, at 73⬚C for 2 minutes, dehydrated in graded ethanol, and air dried. The hybridization mixture contained 4–10 ng of centromeric and 20–40 ng of RB1 gene-specific DNA probe, 10 ␮g of sonicated (200–500 base pairs) unlabeled human placental DNA (Sigma, Germany) in 2 ⫻ SSC containing 50% formamide and 10% dextran sulfate (pH 7). The probe mix was denatured at 73⬚C for 15 minutes and stored on ice until use or was treated according to the manufacturer’s instructions. Ten microliters of denatured probe was applied to the denatured target cells, covered with a cover slip, sealed with rubber cement, and incubated overnight at 37⬚C in a humidified chamber. Slides were washed three times (10 min each) in 55% formamide/2 ⫻ SSC (pH 7) at 45⬚C, once in 2 ⫻ SSC at 45⬚C, and once in 2 ⫻ SSC at RT. To prevent nonspecific antibody from binding to biotinylated probes, the cells were treated in the antibody-diluting buffer solution (4 ⫻ SSC/0.1% Triton X-100/1% bovine serum albumin) at RT for 30 minutes and then immunostained at RT for 60 minutes by using 5 ␮g/mL fluorescence isothiocyanate-labeled avidin (Vector Laboratories). The incubation was followed by 10-minute washes in 4 ⫻ SSC, 4 ⫻ SSC/0.1% Triton-X100, and 4 ⫻ SSC. Directly labeled unbound probes (Spectrum Orangelabeled 13q14) were removed by using the fast postwashing procedure as recommended by the manufacturer: 1 minute in 0.4% NP40/4 ⫻ SSC at 73⬚C, 2 minutes in 2 ⫻ SSC at RT. The cells were air dried and counterstained with DAPI in Vectashield solution (Vector Laboratories). Microscopic Evaluation Slides were scored for the number of hybridization signals in each cell sample by using an epiluminescence microscope (Zeiss Axioplan, Germany) equipped with a 100⫻ oil-immersion objective. At least 200 (when it was possible 400–600) nuclei with intact morphology were examined for each hybridization. The nuclei whose nuclear boundary was broken or torn were considered damaged

3

Mutation of the RB1 Gene

Figure 2 FISH on peripheral lymphocytes obtained from the Figure 1 FISH on retinoblastoma tumor cells. To eliminate the possibility of hybridization error, lymphocytes (treated with hypotonic KCl solution) from a healthy donor were layered as a control on top of the tumor cells (arrow). With the use of RB1 locus-specific probe (specific to 13q14), almost all tumor cells (96%) showed only one fluorescent signal per nucleus. The control cell exhibited two signals per cell.

and were not included in the analysis. Similarly, squashed, smeared, clumped, or overlapped nuclei also were ignored. Each hybridization was accompanied by investigating control normal male lymphocytes.

patient’s blood. One spot per cell for the RB1 gene was observed.

PCR Amplification The DNA was prepared by using standard protocols [20] from the formalin-fixed and paraffin-embedded tumor tissue. The oligonucleotide primers were designed by using the sequence of the coding region of RB1, together with approximately 200 bp of the introns flanking each exon [13, 14]. The oligonucleotides designed to amplify exons 1 through 26, the coding region of exon 27, and the 5⬘ pro-

Figure 3 Immunohistochemical staining of tumor tissue by using an anti-NSE antibody. Formalin-fixed and paraffin-embedded tissue sections were taken from the same part of the tumor as that used for FISH and labeled with an anti-NSE antibody. The majority of the cells were positive for this reaction (appearing red in the photograph), confirming the presence of malignant cells. The nuclei were stained with DAPI (blue).

4

J. Damjanovich et al.

moter region were purchased from Pharmacia (Uppsala, Sweden). Each PCR reaction of a final volume of 50 ␮L contained more then 0.5 ␮g of genomic DNA, 50 pmol each of sense and antisense primers, 2.5 mM MgCl2, 200 ␮M desoxynucleotide triphosphates, 10 mM Tris-HCl (pH 8.4), 50 mM KCl, 0.1% (wt/vol) Triton X-100, and 1.5 U of Taq polymerase (Promega, Madison, WI). The 5⬘ promoter region and exon 1 were amplified by using 10% DMSO. Exons 15 and 16 were amplified together. The reactions were performed on a DNA thermocyler (Hybaid, Teddington, UK). After 5 minutes at 95⬚C, a total of 30–40 cycles were run under the following conditions: denaturation at 95⬚C for 60 seconds, annealing at 55–62⬚C for 60 seconds, and elongation at 72⬚C for 90 seconds, followed by final elongation at 72⬚C for 5 minutes. After PCR, 25 ␮L of the final amplified reaction product was analyzed by electrophoresis on 1.2% agarose gel. SSCP Analysis SSCP analysis was modified from Orita et al. [15]. Because the sensitivity of detection of single-base mismatches by SSCP is inversely proportional to the size of the amplified products [16], the PCR fragments longer than 300 bp were digested with restriction endonucleases [14] (Promega, Madison, WI) for 1 hour at 37⬚C. Amplification product (2 ␮L) was mixed with 28 ␮L of 95% deionized formamide, 10 nM NaOH, 0.05% bromophenol blue, and 0.05% xylene cyanol, incubated at 95⬚C for 2 minutes, and then rapidly chilled on ice. Electrophoresis was done in 7.5% polyacrilamide gel (BioRad, Hercules, CA) with or without 10% glycerol. Gels (16 cm ⫻ 16 cm ⫻ 1 mm) were run for 4 hours at 4⬚C at 300 V in 1 ⫻ TBE buffer. Finally, bands were visualized by the silver-staining method.

RESULTS Interphase Analysis in Normal Peripheral Cells FISH with centromeric DNA probes resulted in specific, intense staining of the centromeres (1, 7, 9, 11, 13/21, 15, 17, 20, X, and Y) in normal cells. Two hybridization signals were detected in about 94% of normal lymphocytes. The 13/21 DNA probe resulted in four hybridization signals in 96% of the control cells. Normal interphase nuclei showed more than 90% of the nuclei with two signals for the RB1 gene.

Figure 4 Immunohistochemical staining for RB protein in formalin-fixed and paraffin-embedded tissue sections of embryonic skin (A) and cartilaginous (B) tissue used as a control (the positive control reactions indicated by the arrows). The formalinfixed and paraffin-embedded tissue sections of the retinoblastoma tumor (C, D) did not show staining for the RB protein; the dark cell layer indicated by an arrow in part C is the retinal pigment epithelium.

FISH to Retinoblastoma Tumor Cells and Peripheral Lymphocytes To determine the copy-number distribution for chromosomes in the highly malignant tumor cells, ten different centromere-specific DNA probes were applied. Approximately 95% of tumor cells showed two hybridization domains for the tested chromosomes. The copy number for chromosomes X and Y was one in about 98% of cells obtained from the tumor. The hybridization of retinoblastoma tumor nuclei with probes for chromosomes 13/21 yielded four clear signals with an incidence rate of 96%. With the use of RB1 locus-specific probe (specific to

13q14), almost all tumor cells (96%) showed only one fluorescent signal (Fig. 1). A similar distribution (1 spot/cell) for the RB1 gene was observed in mononuclear cells from the peripheral blood of the patient (Fig. 2). The proportion of cells with two signals was less then 5% in both cases. In one type of experiment, normal lymphocytes were dropped on top of retinoblastoma cells. After hybridization with the RB1 probe, clear, disomic-copy-number distribution was noted in the normal cells, whereas the tumor cells had one signal per cell. The hybridization error was further minimized by applying an RB1-specific probe from another company (LSI 13 Spectrum Green/LSI21

5

Mutation of the RB1 Gene

Figure 5 PCR products of promoter region and all exons of RB1 from the retinoblastoma tumor sample. Lanes 2– 28 show the results of amplification of promoter (Pro) and exons 1–27 of the retinoblastoma gene. The exact size of the amplimers can be found in Table 1. Lanes 1 and 29 are molecular-weight markers (M) (Sigma, St. Louis, MO).

Spectrum Orange probe from Vysis). With the use of this probe, a signal distribution was observed similar to that heretofore described. FISH was carried out on peripheral lymphocytes from the patient’s parents to determine the origin of the RB1 gene deletion. The copy-number distribution of the RB1 gene in the parents’ mononuclear cells was normal, the majority of cells showing two signals per nucleus. Immunohistochemical Reactions on the Tumor Tissue Because the tumor cells and the peripheral lymphocytes of the patient also showed one signal per cell for the RB1 locus, the immunofluorescent technique was used to identify the ratio of tumor cells in the tissue examined. On tissue sections from the same part of the tumor as that used for FISH, the majority of the cells (93%) were positive for neuron-specific enolase (Fig. 3), confirming that the cells studied were almost exclusively tumor cells. There was no evidence of staining for the RB protein in the paraffin-embedded tissue section of the retinoblastoma tumor, whereas the control tissues showed a positive reaction with the same antibody (Fig. 4). Enzymatic Gene Amplification The PCR results illustrated in Figures 5 and 6 indicate that all PCRs were successful except the amplification of the promoter region, exon 13, exon 15–16, and exon 24 (Fig. 5). The exact size of exon 1 is 307 bp, but the result of this amplification is approximately a 250-bp-length DNA fragment. The expected sizes of the promoter, exon 13, and exon 24 are 570 bp, 570 bp, and 579 bp, respectively.

None of these regions was amplified in the tumor DNA; on the other hand, amplifications of these regions from the control lymphocytes were successful (Fig. 6). SSCP Analysis The PCR products were further analyzed by the SSCP method. All amplimers from the tumor showed the normal bands such as the control amplimers from the healthy lymphocytes. No altered migration was observed (data not shown). DISCUSSION Retinoblastoma is the most common eye tumor of children. The incidence rate of this neoplasia varies between 1:14000 and 1:35000 of live births [21]. Bilateral involvement of the eyes has been observed in 20–40% of the retinoblastomas [21]. Knudson [4, 22] was the first to assume that patients with bilateral retinoblastomas always show a germ-line or inherited mutation. Approximately 80% of unilateral retinoblastomas occur as a result of somatic, nonhereditary mutation [2]. The most common manifestation of the tumor is leukokoria, when the intraocular tumor grows large enough to shine directly off the tumor surface [23]. In unilateral cases, it is an established practice to remove the affected eye if the retinoblastoma tumor infiltrates more than two-thirds of the globe. The genetic background of the unilateral and often isolated cases has always been revealed by different molecular genetic methods. There are several techniques for identifying alterations at the RB1 locus, the most suitable ones being FISH

Table 1 Sizes of different exons of the retinoblastoma gene No. of exon Pro 1 2 3 4 5 6

Amplimer size (bp)

No. of exon

Amplimer size (bp)

No. of exon

Amplimer size (bp)

No. of exon

Amplimer size (bp)

570 307 409 477 445 488 326

7 8 9 10 11 12 13

430 316 316 492 294 465 570

14 15–16 17 18 19 20 21

212 361 315 221 349 350 363

22 23 24 25 26 27

363 420 579 382 394 218

Abbreviation: Pro, promoter.

6

Figure 6 PCR products of the promoter region (Pro), exon 1, exon 13, exon 15–16, and exon 24 of the retinoblastoma gene. Lanes 2–6 are the regions of the retinoblastoma gene obtained from the tumor. Lanes 7–11 are the regions of the retinoblastoma gene obtained from normal lymphocytes. Lanes 1 and 12 are molecular-weight markers (M).

[8], Southern-blot analysis [24], PCR [25], restriction fragment length polymorphism (RFLP) [26], SSCP, and DNA sequencing [6]. In our case, an interphase cytogenetic study was performed by applying DNA-specific probes and using FISH with the intention of analyzing RB1 gene alterations of a unilateral retinoblastoma tumor of a 3-month-old boy. Ten different centrosome-specific DNA probes were used to determine the ploidy pattern of the tumor cells. The majority (more than 95%) had two copies per cell of chromosome 1, 7, 9, 11, 12, 15, and 20, and one signal per cell for chromosomes X and Y, suggesting a stable karyotype for this tumor. The repeated sequence probe, specific for chromosomes 13 and 21, showed four signals in more than 90% of the tumor cells, indicating the presence of both copies of chromosome 13. However, the locus-specific probe recognizing the 13q14 region—where the retinoblastoma susceptibility gene has been mapped—showed only one signal per nucleus. We performed the NSE immune reaction on tissue sections at the site from which the cells for FISH had been taken, because the patient’s peripheral lymphocytes were also monosomic for the RB1 gene by FISH. Almost all cells were positive for NSE. Xu et al. [18] studied the NSE and S-100 protein expression in retinoblastoma tumors, and their results clearly showed that only NSE-positive cells were neoplastic. The S-100 positive ones included both neoplastic and normal retinal elements. Because of their data, the application of NSE was mandatory for the differentiation between lymphocyte contamination, nonneoplastic retinal elements, and retinoblastoma tumor cells. Lohmann et al. [27] analyzed RB1 gene alterations in a number of isolated unilateral retinoblastoma cases. Because no alterations of the RB1 gene were found in the patients’ lymphocytes, the cases were qualified as a result of somatic (i.e., nonhereditary) mutation of the RB1 locus. In two of their cases, the multifocal appearance of the tumor in the affected eye and the development of the tumor in the other eye during the follow-up period suggested the germcell origin of the malignancies without any detectable sign of mutation at the RB1 locus by DNA sequencing [27].

J. Damjanovich et al. It is very important from a prognostic point of view, whether the tumor has been caused by somatic or hereditary or germ-cell mutations. Patients with somatic mutations do not need repeated ophthalmologic controls in their first years of life [28]. Determination of the nature of mutations is required in identifying those family members who have not inherited the dangerous mutations [28]. Patients with retinoblastomas of nonsomatic origin are at high risk of developing a second malignant tumor later in life [29]. In this case, the RB1 gene mutation is associated with the absence of the retinoblastoma nuclear phosphoprotein that regulated G1 progression and functions through its association with different cellular proteins. The absence of the RB1 gene product that is required for tumor suppression can lead to unregulated cell growth [19]. Our results show only one loss at the RB1 locus in the retinoblastoma tumor investigated by FISH. The loss of the first allele indicated by FISH had to be the first one, because it affected both the peripheral lymphocytes and the retinoblastoma tumor cells. However, the parental peripheral lymphocytes did not show this alteration; therefore we can assume that the first deletion might have taken place in one of the parental germ cells. In this case, a second mutation of the RB1 gene induced the so far unilateral intraocular tumor. DNA samples derived from both the tumor and the healthy lymphocytes were amplified with synthetic oligonucleotides belonging to the promoter region and different exons of the RB1 gene. The degree of the degradation was verified by using ␤-globin primer pairs that amplified the specific 536-bp region. The PCR did not yield any amplimers from the tumor, but the exact size of amplimer was observed in the control lymphocytes. The relatively large deletion detected within exon 1 suggests that this alteration could be responsible for loss of function of the RB1 gene. The expected sizes of the promoter region, exon 13, and exon 24 are even longer than the 536-bp length. Because the DNA from the tumor was strongly degraded, the cause of these unsuccessful reactions could be the fragmented template DNA. The amplifications of these regions from the control lymphocytes were successful. The PCR of the exon 15–16 failed, although the expected size of amplimer was much shorter than the 536-bp (361 bp) length, suggesting that some genetic alteration was present within this region. In summary, our data suggest that the 3-month-old patient has a 90% probability of developing a retinoblastoma in his other eye later in life, because he carries a single RB1 defect in probably all of his somatic cells. This research is supported by the National Research Fund (Grant No. F014683 to J. Damjanovich, Grant No. T 022429 to M. Balázs, and Grant No. T20093 to R. Ádány.

REFERENCES 1. Balmer A, Gaillioud C (1983): Retinoblastoma: diagnosis and treatment. Dev Ophthalmol 7:36–100. 2. Hogg A, Bia B, Onadim Z, Cowell JK (1993): Molecular mechanism of oncogenic mutations in tumors from patients with

Mutation of the RB1 Gene

3. 4. 5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

bilateral and unilateral retinoblastoma. Proc Natl Acad Sci U S A 90:7351–7355. Knudson AG (1993): Antioncogenes and human cancer. Proc Natl Acad Sci U S A 90:10914–10921. Knudson AG (1978): Retinoblastoma: a prototypic hereditary neoplasm. Sem Oncol 5:57. Lee WH, Shew JY, Hong FD, Sery TW, Donoso LA, Young LJ, Brookstein R, Lee EY (1987): The retinoblastoma susceptibility gene encodes a nuclear phosphoprotein associated with DNA binding activity. Nature 329:642–645. Toguchida J, McGee TL, Paterson JC, Eagle JR, Tucker S, Yandell DW (1993): Complete genomic sequence of the human retinoblastoma susceptibility gene. Genomics 17:535–543. Liu Z, Song Y, Bia B, Cowell JK (1995): Germline mutations in the RB1 gene in patients with hereditary retinoblastoma. Genes Chromosom Cancer 14:277–284. Pinkel D, Straume T, Gray JW (1986): Cytogenetic analysis using quantitative highly sensitive fluorescence hybridization. Proc Natl Acad Sci U S A 83:2934–2938. Kallioniemi A, Kaliioniemi OP, Waldman FM, Chen LC, Yu LC, Fung YK, Smith HS, Pinkel D, Gray JW (1992): Detection of retinoblastoma gene copy number in metaphase chromosomes and interphase nuclei by fluorescence in situ hybridization. Cytogenet Cell Genet 60:190–193. Dryja TP, Cavenee WK, White R, Rapaport JM, Petersen R, Albert DM, Bruns GA (1984): Homozygosity of chromosome 13 in retinoblastoma. N Engl J Med 310:550–553. Balázs M, Matsumura K, Moore D, Pinkel D, Gray JW, Waldman FM (1995): Karyotypic heterogeneity and its relation to labeling index in interphase breast tumor cells. Cytometry 20:62–73. Balázs M, Mayall BH, Waldman FM (1991): Interphase cytogenetics of a male breast cancer. Cancer Genet Cytogenet 55:243–247. McGee TL, Yandell DW, Dryja TP (1989): Structure and partial genomic sequence of the human retinoblastoma susceptibility gene. Gene 80:119–128. Hogg A (1996): Identification of mutations in the retinoblastoma gene. In: Methods in Molecular Medicine: Molecular Diagnosis of Cancer. FE Cotter, ed. Humana Press Inc., Totowa, NJ, pp. 123–140. Orita M, Suzuki Y, Sekiya T, Hayashi K (1989): Rapid and sensitive detection of point mutations and DNA polymorphysms using the polymerase chain reaction. Genomics 5:874–879. Sheffield VC, Beck JS, Kwitek AE, Sandstrom DW, Stone EM (1993): The sensitivity of single-strand conformation poly-

7 morphism analysis for the detection of single base substitution. Genomics 16:325–332. 17. Motegi T, Kaga M, Yanagawa Y, Kadowaki H, Watanabe K, Inoue A, Komatsu M, Minoda K (1983): A recognizable pattern of the midface of retinoblastoma patients with interstitial deletion of 13q. Hum Genet 64:160–162. 18. Xu KP, Liu SL, Ni C (1995): Immunohistochemical evidence of neuronal and glial differentiation in retinoblastoma. Br J Ophthalmol 79:771–776. 19. Otterson GA, Chen WD, Coxon AY, Khleif SN, Kaye FJ (1994): Incomplete penetrance of familial retinoblastoma linked to germ-line mutations that result in partial loss of RB function. Proc Natl Acad Sci U S A 94:12036–12040. 20. Koshiba M, Ogawa K, Hamazaki S, Sugiyama T, Ogawa O, Kitajima T (1993): The effect of formalin fixation on DNA and the extraction of high-molecular-weight DNA from fixed and embedded tissues. Pathol Res Pract 189:66–72. 21. Judisch GF, Patil SR (1985): In: Contemporary Issues in Ophthalmology. FC Blodi, ed. Churchill Livingstone, New York, pp. 151–173. 22. Cowell JK, Cragg H (1996): Constitutional nonsense germline mutations in the RB1 gene detected in patients with early onset unilateral retinoblastoma. Eur J Cancer 32A:1749–1752. 23. Vogel F (1979): Genetics of retinoblastoma. Hum Genet 52:1–54. 24. Munier F, Spence MA, Pescia G, Balmer A, Gallioud C, Thonney F, van Melle G, Rutz HP (1992): Paternal selection favoring mutant alleles of the retinoblastoma susceptibility gene. Hum Genet 89:508–512. 25. Yandell DW, Campbell TA, Dayton SH, Petersen R, Walton D, Little JB, McConkie-Rosell A, Buckley EG, Dryja TP (1989): Oncogenic point mutations in the human retinoblastoma gene: their application to genetic counseling. New Engl J Med 321:1689–1695. 26. Cavenee WK, Dryja TP, Phillips RA, Benedict WF, Godbout R, Gallie BL, Murphee AL, Strong LC, White RL (1983): Expression of recessive alleles by chromosomal mechanism in retinoblastoma. Nature 305:779–784. 27. Lohmann DR, Gerick M, Brandt B, Oelschlager U, Lorenz B, Passarge E, Horshemke B (1997): Constitutional RB1-gene mutations in patients with isolated unilateral retinoblastoma. Am J Hum Genet 61:282–294. 28. Liu Z, Song Y, Bia B, Cowell JK (1995): Germline mutations in the RB1 gene in patients with hereditary retinoblastoma. Genes Chromosom Cancer 14:277–284. 29. Abramson DH, Ellsworth RM, Kitchin FD, Tung G (1984): Second nonocular tumors in retinoblastoma survivors: are they radiation induced? Ophthalmology 91:1351–1355.