Radiation and Environmental Biophysics

, Volume 50, Issue 1, pp 135–141

Differential effects of genes of the Rb1 signalling pathway on osteosarcoma incidence and latency in alpha-particle irradiated mice

  • Iria Gonzalez-Vasconcellos
  • Tanja Domke
  • Virginija Kuosaite
  • Irene Esposito
  • Bahar Sanli-Bonazzi
  • Michaela Nathrath
  • Michael J. Atkinson
  • Michael Rosemann
Original Paper

DOI: 10.1007/s00411-010-0339-4

Cite this article as:
Gonzalez-Vasconcellos, I., Domke, T., Kuosaite, V. et al. Radiat Environ Biophys (2011) 50: 135. doi:10.1007/s00411-010-0339-4


Osteosarcoma is the most frequent secondary malignancy following radiotherapy of patients with bilateral retinoblastoma. This suggests that the Rb1 tumour suppressor gene might confer genetic susceptibility towards radiation-induced osteosarcoma. To define the contribution of the Rb1 pathway in the multistep process of radiation carcinogenesis, we evaluated somatic allelic changes affecting the Rb1 gene itself as well as its upstream regulator p16 in murine osteosarcoma induced by 227Th incorporation. To distinguish between the contribution of germline predisposition and the effect of a 2-hit allelic loss, two mouse models harbouring heterozygote germline Rb1 and p16 defects were tested for the incidence and latency of osteosarcoma following irradiation. We could show that all tumours arising in BALB/c × CBA/CA hybrid mice (wild-type for Rb1 and for p16) carried a somatic allelic loss of either the Rb1 gene (76.5%) or the p16 gene (59%). In none of the tumours, we found concordant retention of heterozygosity at both loci. Heterozygote knock-out mice for Rb1 exhibit a significant increase in the incidence of osteosarcoma following 227Th incorporation (22/24 in Rb1+/− vs. 2/18 in Rb1+/+, p = 4 × 10−5), without affecting tumour latency. In contrast, heterozygote knock-out mice for p16 had no significant change in tumour incidence, but a pronounced reduction of latency (LT50% = 355 days in p16+/− vs. 445 days in p16+/+, p = 8 × 10−3). These data suggest that Rb1 germline defects influence early steps of radiation osteosarcomagenesis, whereas alterations in p16 mainly affect later stages of tumour promotion and growth.


Carcinogenesis is perhaps the most widely studied late effect following sublethal radiation exposure, but our understanding of the molecular mechanisms that lead from the initial cellular radiation damage to the later arising neoplasm is still incomplete. One of the great challenges that remain to be answered in radiation carcinogenesis is the issue of a potential germline genetic predisposition that could severely modify individual risk. For a few heritable cancer syndromes with tumour aggregation in families or tumour multiplicity in individual patients, the underlying germline defects confer a high penetrance (Fearon 1997; Ponder 1990). Little is known, however, about a concomitant genetic predisposition to spontaneous tumours and susceptibility for secondary neoplasia arising after radiation exposure. Whereas a few genetic syndromes, such as bilateral retinoblastoma, Gorlin-syndrome or familial breast-cancer, suggest such an interrelation (Broeks et al. 2007; Lindor and Greene 1998; Wong et al. 1997; Stavrou et al. 2001), for others, such as adenomatous polyposis syndrome (APS), neurofibromatosis (NF-1) or multiple endocrine neoplasia (MEN1/2) an increased risk to develop cancer following radiation is not yet known.

Sarcomas are prominent secondary malignancies developing after a preceding radiotherapy, with osteosarcomas being the most frequent (Tucker et al. 1987; Huvos and Woodard 1988; Meadows et al. 1980; Pitcher et al. 1994; Mark et al. 1994; Huvos et al. 1985). In an attempt to identify genetic factors that can modulate the individual risk of radiation-induced osteosarcoma, we recently performed genetic mapping studies in mice. We could show that 5 distinct genomic loci in the mouse genome influence osteosarcoma susceptibility following incorporation of the alpha-emitter 227Th (Rosemann et al. 2006). There, the principal susceptibility locus encompassed the Rb1 tumour suppressor gene, whereas 4 additional loci modulated the Rb1 effect in a quantitative manner. The canonical Rb1 function is known to play its key role in cell-cycle regulation, where its phosphorylation state controls the transition of cells from G1 to S phase or the terminal differentiation into G0 (Zhu 2005; Giacinti and Giordano 2006). Other genes upstream of Rb1 code for proteins that transmit extra- or intracellular signals to control the cell-cycle through Rb1 hyper- or hypophosphorylation. Some of those upstream Rb1 regulators are also suspected to confer cancer predisposition, such as p16 and CDK4 for melanoma (Soufir et al. 1998; Harland et al. 2000) or p27kip for neuro-endocrine and prostate cancer (Pellegata et al. 2006; Chang et al. 2004). None of them, however, where shown to modify the risk of radiation-induced osteosarcoma. These findings suggest a potential function of Rb1 in radiation tumourigenesis, or a potential function of it beyond its role as a key cell-cycle regulator. To better understand the molecular mechanism of Rb1 during the process of multi-stage radiation osteosarcomagenesis, we quantified somatic allelic changes in Rb1 and its upstream regulator p16 in tumours of BALB/c × CBA/CA hybrid mice. Based on these retrospective analyses, we used mouse models with targeted Rb1 and p16 germline defects to distinguish the individual contribution of those two genes on the temporal pattern of tumour formation and incidence after 227Th incorporation in bone.

Materials and methods

Mouse breeding and tumour induction

BALB/c × CBA/CA hybrid mice (BCF1) were purchased from Charles River Germany. Heterozygous p16 knock-out mice p16+/− (Sharpless et al. 2001) and homozygous Rb1-LoxP mice Rb1loxP/loxP (Marino et al. 2000) were purchased from the NCI repository of the mouse models of human cancer consortium (MMHCC). Heterozygous transgenic mice expressing Cre recombinase under control of the collagen1a1 promoter (Cre-col) (Dacquin et al. 2002) were purchased from the NIH Mutant Mouse Regional Resource Centre (MMRRC). p16+/− mice were on C57/BL6 background, whereas Rb1-LoxP and collagen 1a1-Cre mice were both on FVB/N inbred background. To induce osteosarcoma, female mice of the following strains were injected with a single dose of 185Bq/g 227Th at the age of 15 weeks: BCF1 (80 animals), C57/BL6 × C57/BL6-p16+/− (70 animals) and FVB/N-CreCol × FVB/N-Rb1-LoxP (42 animals). Preparation, application and dosimetry of the radionuclide 227Th were performed as described previously (Rosemann et al. 2006). Mice were housed 5 to a cage and examined 5 days a week for the development of tumours or other life-threatening conditions. Bone tumours were diagnosed radiologically and confirmed by histological examination after EDTA decalcification. All animal procedures were approved by the state government (file no. RegOB 211-2531-56/06) and carried out in accordance with national animal welfare guidelines.

DNA extraction

DNA from healthy tissue was extracted from 5-mm tail clips. Following overnight lysis at 55°C in buffer containing SDS, Tris-Base, proteinase K and NaCl, proteins, salts and membrane debris were precipitated by adding 2 M NaCl and centrifugation at 12 × 103 rpm for 15 min. Supernatant was collected and DNA precipitated with ethanol. The resulting pellet was washed in 75% ethanol, dried and re-dissolved in ultrapure water yielding a concentration of approximately 300 ng/μl. For subsequent PCR amplification, the concentration was adjusted to approximately 20 ng/μl.

DNA from osteosarcoma tissues was extracted following formalin-fixation, EDTA-decalcification and paraffin-embedding of the tumour samples. Tissue sections of 20 μm were cut on a Leica microtome and mounted on superfrost slides. Sections were dewaxed in xylene and rehydrated in a series from 100 to 10% ethanol. Areas containing predominantly tumour cells were identified using adjacent H&E stained sections and were macro-dissected. DNA was extracted and purified using the DNAeasy mini tissue kit (Quiagen, Hilden Germany). DNA concentration was adjusted to 20 ng/μl.


Alterations in the maternal and paternal allelic ratio of the Rb1 and p16 loci between normal and tumour cells were assayed using a set of polymorphic microsatellite markers (Table 1) following PCR amplification. In brief, PCR was carried out in 96-well plates as described before (Rosemann et al. 2002) using GoTAQ PCR master mix (Promega Inc., Madison, WI) and a GeneAmp 9700 thermocycler (Applied Biosystems, Foster City, CA, USA). Primer sequences and reaction conditions were as recommended by Dietrich et al. 1994. Electrophoresis was done for 90 min on 3% agarose/TBE gels at 50 V/cm.
Table 1

PCR primer sequence for polymorphic markers as used for LOH analysis



5′ Primer

3′ Primer

























Germline genotypes were determined by allele-specific PCRs (Table 2). This way, mice with a heterozygous p16+/− defect can be distinguished from their littermate p16+/+ and heterozygote Rb1-LoxP/CreCol+ can be distinguished from littermate Rb1+/+ and or Rb1-LoxP without the CreCol transgene.
Table 2

PCR primer sequences to genotype germline knockout and transgenic alleles


5′ Primer

3′ Primer













Detection of allelic losses in tumour cells

For quantification of somatic allelic losses of the Rb1 and p16 genes in BCF1 hybrid mice, the relative intensity of the paternal and the maternal alleles were quantified in normal tissue and in tumour DNA. For this purpose, polymorphic microsatellite markers flanking these two genes were PCR amplified and resolved by electrophoresis. The image was electronically captured, and the relative optical density of each allele determined as described previously (Rosemann et al. 2003). A deviation of the allelic ratio of more than 40% compared to the value obtained in matched normal tissue was used as an arbitrary cut-off.


For all statistical calculations, the software-package Statistika 6.0 (StatSoft Inc, Tulsa, OK) was used. Differences in the tumour incidence (number of tumour-bearing mice vs. number of tumour free mice) were tested for significance using Fisher–Yates exact test (two-tailed). Differences in the distribution of tumour latency times were tested for significance using Mann–Whitney’s U-test. Kaplan–Maier survival curves for osteosarcoma development were calculated to account for competing causes of death.


Osteosarcoma induction in BCF1 mice and allelic imbalances

After incorporation of 185Bq 227Th per gram of body weight at the age of 15 weeks, 25 out of 80 female BCF1 mice developed osteosarcoma with latency times between 290 and 766 days (median 553 days, Fig. 1). Tumours originated most frequently from the bones of the extremities (32% in femur, tibia, patella, humerus and ulna), followed by ilium and sacrum (28%), vertebrae (28%) and the skull, including the jaws (12%).
Fig. 1

Kaplan–Meier curve for development of osteosarcoma in female BALB/c × CBA/CA F1 hybrid mice after 227Th injection. 185Bq 227Th (as Thorium citrate) were injected per gram of body weight at the age of 15 weeks. Osteosarcoma formation was monitored and incidence curves calculated with correction for competing causes of death

The Rb1 locus was affected by a loss of one allele, in 13 of 17 available tumours with the maternal BALB/c allele being reduced or lost in 10 cases and the paternal CBA/CA allele reduced or lost in 3 cases (Fig. 2). The p16 locus was affected by allelic imbalance in 10/17 cases, with no preference for either the maternal BALB/c or the paternal CBA/CA. The 4 tumours retaining heterozygosity at Rb1 all showed loss of one p16 allele, so that every tumour demonstrated allelic loss or imbalance of at least one of the two loci.
Fig. 2

Pattern of allelic loss in 17 radiation-induced osteosarcomas found in BALB/c × CBA/CA F1 hybrid mice. Two loci in the mouse genome, encompassing Rb1 on chromosome 14 and encompassing p16 on chromosome 4 are shown. Allelotyping was done for intragenic SNPs of the two genes and for microsatellite markers flanking them proximally and distally. Black: retention of heterozygosity. Grey: loss of maternal BALB/c allele. White: loss of paternal CBA/CA allele

Osteosarcoma induction and latency in mice with a germline Rb1 defect

Out of 42 animals from the Rb1-LoxP x CreCol mating, 13 developed osteosarcoma with a median latency time of 402 days. Eleven of the 24 mice that harbour a deletion of one copy of Rb1 in normal bone (due to the presence of Rb1-loxP and active Cre expression) developed osteosarcoma (median latency 402 days). Of the 18 mice without this pre-existing Rb1 deletion in bone (i.e. absence of Cre expression), only 2 were diagnosed with an osteosarcoma (median latency 392 days). As shown in Fig. 3, the latency period for osteosarcoma induction is similar for both genotypes, but the tumour incidence is significantly different (p = 4 × 10−5 Fisher–Yates exact test).
Fig. 3

Kaplan–Meier curve for development of osteosarcoma in female FVB/N-Rb1LoxP × CreCol mice after 227Th injection. All mice carried heterozygote or homozygote alleles of the conditional Rb1-LoxP allele. Full line: Mice without expression of Cre recombinase. Rb1-LoxP/CreCol(−). Dashed line: Mice expressing Cre recombinase. Rb1-LoxP/CreCol(+)

Osteosarcoma induction and latency in mice with a germline p16 defect

Of a total of 70 animals form the C57/BL6 p16+/− mating, 24 developed osteosarcoma (median latency time 413 days). Of the 36 mice that already harbour a heterozygous p16 defect in the entire germline, 14 developed osteosarcoma (median latency 355 days). Of the remaining 34 mice without a pre-existing p16 germline defect, 10 were diagnosed with an osteosarcoma (median latency 445 days). The tumour incidence in the two genotype groups are statistically not different (p = 0.28, Fisher’s exact test), but they arise much earlier in p16+/− mice in comparison with their wild-type littermates (p = 0.018, Mann–Whitney Test) (Fig. 4).
Fig. 4

Kaplan–Meier curve for development of osteosarcoma in female C57/BL6-p16 k.o. mice after 227Th injection. Littermate mice are either p16+/− heterozygote or p16 wild-type. Full line: Mice inheriting wild-type p16+/+.Dashed line: Mice inheriting p16+/− germline defect. Dotted line: Incidence curve of both genotypes pooled

Note that the incidence curve for wild-type mice from the Rb1-LoxP × CreCol and p16 breedings is different. Here, one has to consider a potential influence of the different genetic background. Obviously, the pure FVB/N strain is less prone to osteosarcoma formation when compared to the C57/BL6 background.


Radiation-associated osteosarcoma was one of the first late complications to be associated with a preceding occupational radiation exposure (Rowland et al. 1978; Fry 1998) or radiotherapy (Huvos and Woodard 1988; Huvos et al. 1985; Mark et al. 1994; Meadows et al. 1980; Pitcher et al. 1994; Tucker et al. 1987). The excess risk of radiotherapy-associated secondary bone tumours in patients with bilateral retinoblastoma was estimated as 36.7 when compared to 2.7 in patient without the congenital condition (Wong et al. 1997). This implies that the germline mutation, which in the first instance predisposes for retinoblastoma, also has a promoting effect on the development of osteosarcoma following ionizing radiation. The mutated gene responsible for these effects was identified to be Rb1, which codes for p105-Rb, a regulatory protein of the cell cycle (Giacinti and Giordano 2006; Zhu 2005). The Rb1 pathway consists of upstream elements that control p105-Rb phosphorylation and downstream effectors that transduce the signal further to induce S phase-related genes. One of the major upstream regulatory elements of Rb1 is the CDK-inhibitor p16, which, like Rb1, can function as a tumour suppressor gene (Kamb et al. 1994; Lukas et al. 1995). An effect of the later gene to increase risk of radiation-associated human osteosarcoma was not reported in the literature, but might have just escaped detection due to the small number and genetic heterogeneity of cases. Considering the close functional relationship of both genes, we wanted to know whether a germline defect in p16 can cause an increased susceptibility for radiation-induced osteosarcoma in a similar way to Rb1. We have explored these factors in genetically defined mouse models irradiated by injection of a bone-seeking alpha emitter. We reported hat QTL-mapping of BALB/c × CBA/CA backcrossed mice irradiated in this way showed a strong susceptibility-modifying effect from allelic variants of the Rb1 locus, but none from the p16 locus (Rosemann et al. 2006). The reason being an increase in Rb1 gene expression on the BALB/c allele that leads to the preferential loss on the tumour (unpublished data).

In this study, we asked whether the two genes might be affected by somatic allelic changes during tumourigenesis, as proposed by Knudson’s two-hit model (Knudson 1971). It appears that these tumour-specific allelic changes of Rb1 and p16 are complementary, i.e. retention of the heterozygous state at both loci was mutually exclusive. The absence of a concordant normal state of both Rb1 and p16 is unlikely assuming they arise randomly during tumourigenesis (a concordant heterozygosity by chance would be expected in only 1–2 cases). The only reasonable explanation for the observed discordant allelic imbalances affecting Rb1 and p16 is that both genes are equally important in preventing osteosarcoma formation following radiation. In other words, an allelic loss of either is essential, albeit not sufficient, to drive carcinogenesis. It is interesting to note that in spontaneous human osteosarcoma both Rb1 and p16 are also altered, either by loss of heterozygosity (Rb1) or by an epigenetic downregulation (p16) (Little and Wainwright 1995; Merlo et al. 1995). Both observations support the notion that Rb1 and p16 share a common pathway and a somatic alteration of either of the two genes has a similar promotional effect during tumour growth.

This picture is supported by the canonical function of the Rb pathway in cell-cycle regulation. Here, both p16 and Rb1 act in a serial manner, and both function as negative regulators of the cell-cycle transition from G1 to S phase. Whereas hypophosphorylated Rb1 protein can bind and antagonize S phase promoting transcription factors, thereby keeping cells at G1arrest, the p16 protein is an inhibitor of the Rb1 phosphorylation and hence keeps Rb1 in its active and growth-suppressing state (Lukas et al. 1995).

This close functional relationship between Rb1 and p16 in the growth control of normal cells, and consequently in the pattern of their somatic changes during tumourigenesis, raises the question of whether a similar link between both genes can be found in the process of congenital tumour susceptibility. Here, gene variants or mutations that are present in normal tissue are assumed to mediate the risk of a malignant transformation of cells. Several such germline conditions are known for spontaneous cancer, and a few were also shown to have a strong influence on the individual risk to develop cancer following radiation exposure (Lindor and Greene 1998; Stavrou et al. 2001; Wong et al. 1997). In case of osteosarcoma, Rb1 germline mutations leading to amino acid changes or to a truncated protein are present in some, but few patients with osteosarcoma (less than 3%), remarkably those who during childhood experienced bilateral retinoblastoma and underwent local radiotherapy (Wong et al. 1997). This demonstrates an interaction between a pre-existing genetic predisposition and the genotoxicity of radiation in the process of osteosarcoma development. Germline defects in p16 predispose to cancer as well, but the observed tumour types do not included osteosarcoma or other radiation-induced cancer, but mainly melanoma of the skin and pancreas carcinoma (Harland et al. 2000; Soufir et al. 1998). Whereas a concordant predisposition to osteosarcoma could have hardly been overseen in p16 mutation carriers, the apparent absence of increased risk of radiation-associated cancer might have been due to the fact that p16 mutation carriers rarely undergoing radiotherapy.

To evaluate whether a p16 germline defect influences radiation carcinogenesis similar to an Rb1 defect, we used two different mouse lines with targeted gene defects in either of the two genes and challenged them with an osteosarcoma inducing dose of 227Th. In both cases, the osteosarcoma development in animals heterozygous for the gene defects was significantly different when compared to their wild-type littermates. The pattern of osteosarcoma development after irradiation, however, was markedly different in the two mouse lines: Whereas the Rb1 defect had a clear and strong effect by increasing the osteosarcoma incidence by a factor of more than 4, p16 did not increased tumour incidence significantly. Instead, the tumours grew earlier in p16 knock-out mice, implying that this gene influenced the biology of tumour cells and consequently tumour progression rather than the susceptibility of the normal tissue to undergo malignant transformation. In the context of a multi-step model of tumourigenesis, a pre-existing germline defect of the Rb1 gene might be important in directly influencing how and to what extent an irradiated normal cell can tolerate or repair the induced lesion to its genome. The canonical Rb pathway is involved in maintaining genome integrity by its potential to block the cell-cycle progression at the G1/S boundary in cells carrying unrepaired DNA breaks (Giacinti and Giordano 2006; Zhu 2005). In this context, it is unclear why a defect in p16, although upstream of Rb1 and equally important for cell-cycle regulation, does not increase tumour formation after radiation exposure. It is therefore proposed that an as yet unknown second function of Rb1, independent from the canonical Rb pathway, is specifically involved in the maintenance of genetic stability, as was also recently shown by others (Manning et al. 2010). This observation suggests that the increased susceptibility for radiation-induced osteosarcoma by Rb1 germline mutations in man or a targeted heterozygote Rb1 knock-out in mice is caused by a not yet defined function of p105-Rb in maintaining genomic stability.


Compared to the congenital predisposition for radiation-induced osteosarcoma caused by Rb1 germline defects, a similar defect in p16 does not increase susceptibility. Instead, it causes a significant reduction of tumour latency after radiation exposure. It is thus concluded that p16 is involved in later stages of tumour promotion, rather than in initiation of malignant transformation.


This work was supported by BMBF and BMU via Kompetenzverbund Strahlenforschung (KVSF) grant 03NUK007 and FP6 EU contract “Risc-Rad”.

Copyright information

© Springer-Verlag 2010

Authors and Affiliations

  • Iria Gonzalez-Vasconcellos
    • 1
  • Tanja Domke
    • 1
  • Virginija Kuosaite
    • 2
  • Irene Esposito
    • 2
  • Bahar Sanli-Bonazzi
    • 1
  • Michaela Nathrath
    • 3
  • Michael J. Atkinson
    • 1
  • Michael Rosemann
    • 1
  1. 1.Institute of Radiation Biology, Helmholtz Zentrum MünchenNeuherbergGermany
  2. 2.Institute of Pathology, Helmholtz Zentrum MünchenNeuherbergGermany
  3. 3.Clinical Cooperation Group Osteosarcoma, Helmholtz Zentrum München and TU MünchenNeuherbergGermany

Personalised recommendations