Second, we have develop a comprehensive database of normal tissue and irradiated tissue by analyzing different organs in each mouse. This database will become a public resource.

Third, our studies will identify genetic predisposition to radiation associated cancer, and may thus help protect humans who have the greatest risk of cancer.

Fourth, our studies have provided compelling evidence of radiation action via non-targeted effects, which may inform how radiation risks are estimated and managed prospectively.

The goal of Project 1 was to determine the contribution of NTE as a function of radiation quality on breast cancer. We used a radiation chimera mammary model to discern the effects of radiation-induced microenvironment from DNA damage and potential mutation. In this model we first surgically cleared the mammary gland of endogenous epithelium. Then the mouse was irradiated and the fat-pad subsequently orthotopically transplanted with un-irradiated, non-malignant epithelial cells primed for malignant transformation by Trp53 deletion that were evaluated for tumors for 600 days.

A paper published in Cancer Research in 2014 reported results of studies in which mice were irradiated with either 100cGy of gamma-radiation or 11, 30, or 81 cGy from 350MeV/amu Si particles compared to sham-irradiated controls prior to transplantation with Trp53 null mammary fragments. Trp53 null tumors arising in mice densely irradiated had a shorter median time to appearance and grew faster once detected compared to those in sham-irradiated or gamma-irradiated mice. Tumors were further classified by markers keratin 8/18 (K18, KRT18), keratin 14 (K14, KRT18), and estrogen receptor (ER, ESR1) and expression profiling. Most tumors arising in sham-irradiated hosts contained both K18 and K14 positive cells (K14/18) while most tumor arising in irradiated hosts were K18 alone. K14/18 tumors were predominantly ER positive while K18 tumors were predominantly ER negative. Although K18 tumors tended to grow faster and be more metastatic than K14/18 tumors, K18 tumors in Si particle-irradiated mice grew significantly larger compared to controls and were more metastatic compared to sham-irradiated mice. The expression profile that distinguished K18 tumors arising in particle-irradiated compared sham-irradiated mice was enriched in mammary stem cell, stroma, and Notch signaling genes. These data suggest that densely ionizing radiation carcinogenic effects mediated by the microenvironment elicit more aggressive tumors compared to tumors arising in sham-irradiated hosts.

To compare the contribution of targeted and NTE, we irradiated mice after Trp53 null outgrowth was established (i.e., epithelium was exposed, i.e., targeted and NTE) and the other group was irradiated before transplantation (i.e., NTE only, epithelium was not exposed). Tumors in mice irradiated with densely ionizing 600 MeV/amu Fe charged particles or sparsely ionizing gamma-ray radiation were compared to those in contemporaneous sham-irradiated controls. Irradiation of the in situ mammary outgrowth with either radiation quality decreased Trp53 null tumor latency and increased tumor frequency a year after. Mammary tumor latency arising from non-irradiated Trp53 null outgrowths in irradiated hosts was also decreased indicating that this effect was mediated by the microenvironment. Tumors arising from densely ionizing irradiated outgrowths gave rise to aggressive tumors with increased growth rate compared to sham-irradiated mice. Importantly, increased tumor aggressiveness was also observed in tumors arising from irradiated hosts, suggesting that this effect was mediated by NTE. Only densely ionizing radiation affected tumor markers of differentiation in both models. The data from these studies supports the conclusion that tumor aggressiveness is indirectly promoted by NTE and that NTE elicited by HZE particles results in more aggressive tumors. The RNA of these tumors has been sequenced. Notably, preliminary analysis using unsupervised clustering separated tumors arising from irradiated epithelium in irradiated hosts (i.e., targeted) from those that arose in irradiated mice. These data are in preparation for publication.

We also evaluated the effect of age in the radiation chimera-experiment; mice were irradiated at 10 months, simulating the average age for astronauts. Notably high LET exposure of aged mice is much more efficient inducing tumorigenesis than low LET. Cancer incidence in humans increases exponentially with age, with 75% of newly diagnosed cases occurring in susceptible populations aged 55 years or older. The aging process has been shown to be associated with increased levels of chronic inflammation, which are thought to contribute to many age-associated diseases, including cancer, and increased serum levels of IL-6 have been reported in older individuals. Interestingly, A-bomb survivors have significant increases in serum IL-6 levels that are still detectable five decades after exposure. These findings suggested to us the possibility that radiation carcinogenesis, at least in part, is mediated by induction of a pro-inflammatory environment similar to the aging process. If so, we predicted that radiation NTE would be epistatic with age.

To test this, mice were irradiated at 10 months of age, comparable to the middle age of astronauts during space-flight missions to test the hypothesis that NTE overlap with the biology that increases cancer incidence in aged versus young people. The inguinal mammary glands of 200 mice were cleared at 3 weeks of age and the mice housed under standard conditions. Hosts were irradiated at 10 months of age with 10, 50, or 100 cGy gamma-radiation or 350 MeV/amu Ar or Si, or 600 MeV/amu Fe at a fluence of 1 particle per 102 mm, and transplanted 3 days later with Trp53 null epithelium. Tumorigenesis was monitored for 600 days. As expected due to decreased ovarian hormones in aged mice, fewer mammary outgrowths were obtained in aged mice; outgrowth efficiency is >90% in young mice mammary but ~50% in aged mice. This limited the ability to compare dose or particles individually. However, we were able to compare tumors arising in sham-irradiated mice to gamma-irradiated hosts or HZE-irradiated hosts. Age and radiation NTE were not epistatic as both low- and high-LET irradiation of aged hosts significantly increased tumor incidence and tumor growth rate. Moreover, this experiment revealed a strong radiation quality effect in which HZE NTE was considerably more effective than gamma-radiation.

Expression profiling of these tumors indicate that the top ranked (p>0.00001) processes that distinguish tumors from aged mice irradiated with HZE particles from age-matched, sham-irradiated tumors were inflammation (70 genes), immune cell trafficking (79 genes), and cancer (271 genes). Among the 20 most significant upstream regulators ranked by z-score (p<0.05), the majority are pro-inflammatory factors such as IL-1ß, TNFalpha, and IL6. Consistent with our previous findings, TGFß1 is predicted by IPA to also be a significant regulator in cancers from both HZE and gamma-irradiated host. We speculate that gamma- and HZE-irradiated microenvironments might evoke common mediators, but their magnitude or persistence might be different. Among the top-ranked upregulated molecules are COX2 and granzyme H. We validated this signature using relative COX2 transcript abundance and protein expression, which support high levels of COX-2 in tumors arising in HZE-irradiated mice. These data are being prepared for publication (Ouyang et al.).

Together these data lead us to postulate that a high degree of immunosuppressive inflammation promotes these aggressive tumors. To test this idea, we used immunostaining to evaluate lymphoid and myeloid cellular subsets within tumors. Most strikingly, tumors arising in HZE-irradiated aged hosts have significantly (p<0.05) more myeloid derived suppressor cells (MDSC) and Treg cells compared to tumors arising in age-matched, sham-irradiated hosts. Treg frequency inversely correlates with tumor growth rate and MDSC is positively correlated with Cox2 staining, leading to the new hypothesis that HZE radiation exposure in mid-life is strongly immunosuppressive.

Project 2, directed by Dr. Allan Balmain at UCSF, had the ultimate goal of identifying the genetics associated with susceptibility to cancer risk from high LET radiation. The strategies employed were based on systems genetics approaches to identify germ line polymorphisms and somatic genetic changes associated with tumor development. This project focuses heavily on genetic analysis of radiation-induced tumors. A paper recently published in Molecular Carcinogenesis describes the phenotype of germline deletion of a small N-terminal proline-rich region of Trp53 in an in vivo mouse model (Trp53 deltaP), which causes no spontaneous tumors; a manuscript on this model was published in 2015 by the group. Upon exposure of the Trp53 deltaP mouse on the original mixed background of 129/Sv/C57BL/6 to either low and high LET radiation, a wide range of solid tissue tumors develops, including those of the liver, lung, and kidney. This is in stark contrast to the tumor spectrum in the Trp53 null (-/-) mouse irradiated with 4 Gy low LET, which generates mainly lymphomas and sarcomas. The Trp53 deltaP mouse therefore represents a sensitive model to study radiation-induced tumors similar to those humans.

Interestingly the irradiated tumor spectrum of both Trp53 deltaP and Trp53+/- mice in the FVB/N background falls into three main tumor types: thymic lymphomas, mammary tumors, and skin carcinomas. It should be noted that lymphomas are often observed post-radiation in mice of compromised Trp53 function, particularly after exposure to high radiation doses (4 Gy). After exposure to a low dose (50cGy) of either low or high LET radiation, we observed increased frequency of mammary tumors and skin carcinomas in both genotypes. Although the tumor spectrum is similar between mice exposed to high or low LET radiation, there was an increased incidence of mammary tumors following high LET radiation.

Of note the tumor spectrum in 129/Sv Trp53+/- mice was markedly different from FVB/N. Regardless of radiation quality there was a high incidence of thymic lymphomas. A wide range of tumors of varying types with no distinguishable pattern but no skin carcinomas were observed. A small number of mammary tumors developed following high LET radiation (n=3), whilst none were observed in the gamma-irradiated group. These differences further illustrate the influence of background strain upon tumor latency, susceptibility, and spectrum following exposure to high or low LET radiation. These data demonstrate that FVB/N mice are particularly susceptible to radiation-induced mammary tumor development. High LET radiation was also more potent at inducing a higher incidence of mammary tumors with a shorter latency compared to gamma-radiation (p=0.0004 CoxPH).

Little is known about the mechanisms by which high LET induces carcinogenesis. Exposing mice of abrogated Trp53 function to either high or low LET radiation provided us with a unique opportunity to compare and contrast the genetic aberrations and potential radiation-quality specific signatures in tumors. Next generation sequencing offers a powerful tool to gain a deeper understanding of the complex mutational processes driving radiation-induced tumorigenesis. In collaboration with Dr. David Adams at Sanger Institute, Cambridge, exome sequencing was performed on 60 tumor samples, along with 10 control samples. We have sequenced the most frequently observed tumor types, (thymic lymphomas, mammary tumors, and skin carcinomas), in Trp53 deltaP and Trp53+/- on FVB/N or 129/Sv/C57BL/6 backgrounds induced by either Fe-ion or gamma radiation.

Project 3, led by Dr. S. V. Costes and Dr. Mao, undertook to use computational modeling of experimental data. A manuscript was published in Radiation Research that predicts RBE for cell death based on experimental measurements done with X-ray. Briefly, Linear-quadratic models have been used for decades to interpret cell survival after exposure to ionizing radiation. However, parameters for these models change with LET, forcing investigators to measure experimentally cell survival for any LET of interest. In contrast to the current paradigm, we hypothesize that double strand breaks (DSBs) are moved into repair domains and consequently merge into clusters. Overall, this work suggests that microdosimetric properties of ion tracks at the sub-micron level are sufficient to explain both RIF data and survival curves for any LET, similarly to the Local Effect Model assumption.

The use of agent-based modeling (ABM) to estimate the contribution of stem cells to carcinogenesis published in 2013 in collaboration with Project 1, was extended to allow integration of the multi-stage clonal expansion model with NTE agent-based approaches. The goal was to simulate tumor data in silico by assuming that tumors arise via successive DNA mutation and that promotion is influenced by non-targeted effects based on the experimental data in Project 1. One challenge in generating tumor in silico is the duration of simulations using our current ABM, therefore we adopted a simpler computer model (automata) that allows cells to be tracked faster over a 60 year simulation (from age 20 to 80 years old) to implement the multi-stage clonal expansion model to simulate tumor incidence arising spontaneously in human population due to random mutations using these automata. Each simulated person is a monolayer of 4 million cells (referred as “tissue”), where cells divide only when neighboring cells have died. Parameter sweeps were done so that simulations predicted cancer incidence matching a normal human population.

Second, we will develop a comprehensive database of normal tissue and irradiated tissue by analyzing different organs in each mouse. This database will become a public resource.

Third, our studies will identify genetic predisposition to radiation associated cancer, and may thus help protect humans who have the greatest risk of cancer.

The goal of Project 1 is to determine the contribution of NTE as a function of radiation quality on breast cancer. Aim 1.1 uses a mammary chimera model in which wildtype Balb/c mice are transplanted orthotopically with syngeneic Trp53 null mammary epithelium. Tumorigenesis was compared in mice irradiated with different particles before transplantation (i.e. the radiation chimera, in which epithelium is not exposed), irradiation of intact mammary glands (i.e. the genetic chimera, in which a Trp53 null epithelial outgrowth is irradiated in situ), and in aged mice in which 10 month old mice were irradiated before transplantation. Aim 1.2 investigates the inflammatory signals induced by radiation. Aim 1.3 uses systems biology to integrate data from different models and types to provide a better mechanistic understanding of underlying biology.

It is unknown to what extent the carcinogenic potential as a function of radiation quality is mediated by differential effects on the irradiated microenvironment. To address this issue, we employed a radiation chimera mammary model to separate radiation-induced microenvironment from radiation-induced mammary epithelial DNA damage and potential mutation. Our current analyses show that NTE do contribute to carcinogenesis by affecting tumor latency, growth rate, and characteristics. Tumors affected only by radiation NTE, i.e. arising in irradiated mice, were far more aggressive than those from non-irradiated mice. In contrast tumors arising from irradiated Trp53 epithelium were distinct in that neither ER-negative tumors nor metastases were increases. We conclude that aggressive tumors are promoted by radiation NTE host responses such as oscillatory inflammation.

We have submitted a manuscript that shows that Trp53 tumors arising in Si-irradiated hosts are enriched in a K18-positive, ER-negative subtype but these tumors are different from similar tumors that arose in sham-irradiated mice. These analyses show that NTE do contribute to carcinogenesis by affecting tumor latency, tumor growth rate, and tumor features. The biological behavior and features of tumors arising from un-irradiated Trp53 null mammary epithelium transplanted to sham-irradiated mice were compared to tumors arising in mice contemporaneously irradiated with 350 MeV/amu Si particles or a reference dose of sparsely ionizing gamma-radiation. Tumors arising in mice exposed to densely ionizing radiation grew significantly faster, were more likely to metastasize, and could be genomically separated from tumors with the same markers arising in sham-irradiated mice. These data suggest that the greater carcinogenic effect of densely ionizing radiation is mediated in part by the microenvironment, which elicits more aggressive tumors compared to similar tumors arising in sham-irradiated hosts.

Although we do not observe a robust LET and fluence dependence, given our evolving understanding of the NTE dose-dependence, it may well be that thresholds rather than proportional responses may be more critical. For example, it may be that each NTE has a distinct threshold and that these change as a function of LET. The effect on tumor type (e.g. ER statues) appears to be the most sensitive index in the radiation chimera model. A dose threshold for affecting tumor ER status was exceeded by 10 cGy low LET radiation or with a fluence of 1 particle/100 microm2 of high LET particles. The lack of classical dose dependence is challenging in designing further animal experiments that are lengthy and costly.

Project 2 has the ultimate goal of identifying the genes and pathways associated with susceptibility to cancer risk from high LET radiation. The strategies employed are based on previous experience in using systems genetics approaches to identify germ line polymorphisms and somatic genetic changes associated with tumor development induced by chemicals or ionizing (low LET) radiation. This project is therefore focused heavily on genetic analysis of radiation-induced tumors, and as such complements the goals of Project 1 on analysis of the effects of radiation exposure on the normal or tumor microenvironment.

The tumor suppressor Trp53, which is regularly mutated in cancers, is often referred to as the guardian of the genome due to the multiplicity of anti-proliferative effects it can initiate to maintain genomic integrity under genotoxic stress conditions. To investigate the impact of both low and high LET-radiation upon carcinogenesis we used the Trp53 deltaP mouse model created by our collaborator Geoff Wahl. Germline deletion of a small N-terminal proline-rich region of Trp53 in an in vivo mouse model (Trp53 deltaP) causes no spontaneous tumors. Interestingly, upon exposure of the Trp53 deltaP mouse to both low and high LET (linear energy transfer) radiation, a wide range of solid tissue tumors develops, including those of the liver, lung, and kidney. This is in stark contrast to the tumor spectrum in the Trp53 null (-/-) mouse post-radiation, which comprises mainly lymphomas and sarcomas. The Trp53 deltaP mouse therefore represents an exquisitely radiation-sensitive model to study radiation-induced tumors similar to those seen in the human population. Using a systems genetics approach as described above, we will generate an integrated network view of the factors which influence the susceptibility of certain tissues to the effects of both low and high LET radiation. Furthermore, this analysis will enable the identification of specific genes and pathways which are associated with the tumor-inducing effect of radiation. These data, in combination with next-generation sequencing data, enabling the identification of recurrent genetic aberrations occurring during tumorigenesis will provide insight to identifying potential genetic susceptibilities for future assessment in humans.

In previous experiments we have observed a wide range of tumor types in various organs following irradiation of the Trp53 deltaP mouse on the original mixed background of 129/Sv/C57BL/6. Interestingly the irradiated tumor spectrum in the FVB/N background of both Trp53 deltaP and Trp53+/- mice falls into three main tumor types, thymic lymphomas, mammary tumors, and skin carcinomas. It should be noted that lymphomas are often observed post-radiation in mice of compromised Trp53 function, particularly after exposure to relatively high doses of ionizing radiation (4Gy). After exposure to lower doses (50cGy) of either low or high LET radiation, we observed an increased frequency of mammary tumors and skin carcinomas also in Trp53+/- mice. Although the tumor spectrum is similar between mice exposed to high or low LET radiation, there is an increased incidence of mammary tumors following high LET radiation.

Of note the tumor spectrum in 129/Sv Trp53+/- mice was markedly different from FVB/N. Regardless of radiation quality there was a high incidence of thymic lymphomas, and a wide range of tumors of varying types with no distinguishable pattern and also no skin carcinomas were observed. A very small subset of mammary tumors developed following high LET radiation (n=3), whilst none were observed in the gamma irradiated group. These differences further illustrate the influence of background strain upon tumor latency, susceptibility and spectrum following exposure to high or low LET radiation. These data demonstrate that FVB/N mice are particularly susceptible to radiation-induced mammary tumor development. High LET radiation is also more potent at inducing a higher incidence of mammary tumors, and shorter latency compared to gamma radiation (p=0.0004 CoxPH).

Little is known about the mechanisms by which high LET induces carcinogenesis. Exposing mice of abrogated Trp53 function to either high or low LET radiation provides us with a unique opportunity to compare and contrast the genetic aberrations and potential radiation-quality specific signatures in tumors. Next generation sequencing offers a powerful tool to gain a deeper understanding of the complex mutational processes driving radiation-induced tumorigenesis. In collaboration with Dr. David Adams at Sanger Institute, Cambridge, exome sequencing has been performed on 60 tumor samples, along with 10 control samples. We sequenced the most frequently observed tumor types, (thymic lymphomas, mammary tumors and skin carcinomas), in Trp53 deltaP and Trp53+/- on FVB/N or 129/Sv/C57BL/6 backgrounds induced by either Fe-ion or gamma radiation.

Project 3

Predicting cell survival RBE for any HZE using microdosimetry and a DNA damage clustering model. A manuscript has been submitted that predicts RBE for cell death based on experimental measurements done with X-ray. Briefly, Linear-quadratic models have been used for decades to interpret cell survival after exposure to ionizing radiation. However, parameters for these models change with LET, forcing investigators to measure experimentally cell survival for any LET of interest. In contrast to the current paradigm, we hypothesize that DSBs are moved into repair domains and consequently can be merged into clusters. Experimental work characterizing the dose dependence of radiation-induced foci (RIF) from X-ray in nonmalignant human mammary epithelial cells (MCF10A) is used here to validate a DSB clustering model. Cell death probability is computed as a decreasing exponential proportional to the number of DSBs and all combinations of pairs of DSBs that can occur in each cluster. This leads to two cell death coefficients from all DSBs and paired DSBs respectively. These coefficients can be inferred via computer simulations, by fitting survival curves of MCF10A exposed to X-ray. The clustering model that led to the best predictions assumes that the nucleus is divided into an array of regularly spaced repair domains. This model was then tested to predict the dose response of RIF and cell death after exposure to densely ionizing particles, as opposed to X-ray. Predicted RIF/µm had an average relative error of 12% when compared to experimental values covering a range of LET between 30 and 350 keV/µm for three different ions. Cell survival RBE predictions for MCF10A exposed to any LET remain to be verified experimentally. However, the predicted values and the LET dependence we report here match previous experimental results on similar cell types. With such an approach, parameters derived experimentally on X-ray are sufficient to predict the relative biological effectiveness of high-LET radiation by simply including microdosimetric properties of ion tracks. Overall, this work suggests that microdosimetric properties of ion tracks at the sub-micron level are sufficient to explain both RIF data and survival curves for any LET, similarly to the Local Effect Model (LEM) assumption. On the other hand, high-LET death mechanism does not have to infer linear-quadratic dose formalism as done in the LEM. Our model is therefore an alternative to previous approaches by providing a testable biological mechanism (i.e. RIF).This approach opens the door to a unified model to predict risk to cosmic rays for astronauts and we plan to extend it to also model RBE prediction for genomic instability.

Current development: integrating multi-stage clonal expansion model with NTE agent-based approaches. Our goal for the grant is to simulate tumor data in silico by assuming that tumors rises via successive DNA mutation and that promotion are influenced by non-targeted effects. One challenge in generating tumor in silico is the duration of simulations using our current ABM. We have switched to a simpler computer model (automata) that allows cells to be tracked faster over a 60 year simulation (from age 20 to 80 years old). We have currently implemented the multi-stage clonal expansion model (MSCE) to simulate tumor incidence arising spontaneously in human population due to random mutations using these automata. Each simulated person is a monolayer of 4 million cells (referred as “tissue”), where cells divide only when neighboring cells have died. Parameter sweep was done so that simulations predicted cancer incidence matching normal human population (~50% depending on sources). Our current parameters values are as follow: cell death frequency (beta term in MSCE) was set to 0.008, mutation frequency was set to 3E-8 per division, cell cycle time is set to 24 hours (experimental value) and we assume tumor arises via two events: initiation and conversion (v and µ term respectively in Two-Stage CE). The model is set to allow more stages before tumor arises if necessary (i.e. MSCE). Once a cell has become fully transformed, it can invade neighboring cells with an adjustable parameter for invasion efficiency. We plan to include radiation effects at two levels: targeted effect (as described in previous section) generating transient increase of mutation rate and non-targeted effects by modifying transformation rates transiently (transient time matching measurement made in this work and others).

Second, we will develop a comprehensive data base of normal tissue and irradiated tissue by analyzing different organs in each mouse. This data base will become a public resource.

Third, our studies will identify genetic predisposition to radiation associated cancer, and may thus help protect humans who have the greatest risk of cancer.

The goal of Project 1 is to determine the contribution of NTE as a function of radiation quality on breast cancer. Aim 1.1 uses a mammary chimera model in which wildtype Balb/c mice are transplanted orthotopically with syngeneic Trp53 null mammary epithelium. Tumorigenesis was compared in mice irradiated with different particles before transplantation (i.e. the radiation chimera), in which epithelium is not exposed), irradiation of intact mammary glands (i.e. the genetic chimera, in which a Trp53 null epithelial outgrowth is irradiated in situ), and in aged mice in which 10 month old mice were irradiated before transplantation. Aim 1.2 investigates the inflammatory signals induced by radiation. Aim 1.3 uses systems biology to integrate data from different models and types to provide a better mechanistic understanding of underlying biology. NOTE: The Barcellos-Hoff and Demaria labs at NYU were heavily damaged by Hurricane Sandy on October 29th, 2012. As a consequence, all work ceased for 3 full months while we were relocated to a new laboratory in January. Several reagents were compromised by loss of power, in particular the Trp53 null tissue fragments needed for the transplantation experiments. As a consequence we are unable to conduct the low dose rate experiments planned for this spring and summer.

Our current analyses show that NTE do contribute to carcinogenesis by affecting tumor latency, growth rate, and characteristics. Tumors affected only by radiation NTE, i.e. arising in irradiated mice, were far more aggressive than those from non-irradiated mice. In contrast tumors arising from irradiated Trp53 epithelium were distinct in that neither ER-negative tumors nor metastases were increases. We conclude that aggressive tumors are promoted by radiation NTE host responses such as oscillatory inflammation.

We have not observed a robust LET dependence but given our evolving understanding of the NTE dose-dependence, it may well be that thresholds rather than proportional responses may be more critical. For example, it may be that each NTE has a distinct threshold and that these change as a function of LET. The effect on tumor type (e.g. ER statues) appears to be the most sensitive index in the radiation chimera model. This threshold affecting tumor type was exceeded with doses of 10 cGy (Nguyen et al., 2011) and with a fluence of 1 particle/100 microm2 of high LET particles. The lack of classical dose dependence is challenge in the design of animal experiments.

Gene expression profiles of tumors that arose in irradiated mice are distinct from those that arose in naïve hosts. We asked whether expression metaprofiles could discern radiation-preceded human cancer or be informative in sporadic breast cancers. Human orthologs of the host irradiation metaprofile discriminated between radiation-preceded and sporadic human thyroid cancers. Detection of radiation-preceded human cancer by the irradiated host metaprofile raises possibilities of assessing human cancer etiology. This paper was published in Clinical Cancer Research in February as was accompanied by a perspective written by Dr. Barcellos-Hoff on the role of tissue microenvironment in determining aggressive tumor types.

Project 2 has the ultimate goal of identifying the genes and pathways associated with susceptibility to cancer risk from high LET radiation. The strategies employed are based on previous experience in using systems genetics approaches to identify germ line polymorphisms and somatic genetic changes associated with tumor development induced by chemicals or ionizing (low LET) radiation. This project is therefore focused heavily on genetic analysis of radiation-induced tumors, and as such complements the goals of Project 1 on analysis of the effects of radiation exposure on the normal or tumor microenvironment.

The tumor suppressor Trp53, which is regularly mutated in cancers, is often referred to as the guardian of the genome due to the multiplicity of anti-proliferative effects it can initiate to maintain genomic integrity under genotoxic stress conditions. Germline deletion of a small N-terminal proline-rich region of Trp53 in an in vivo mouse model (Trp53 delta P) causes no spontaneous tumors; yet, upon exposure of the Trp53 delta P mouse to both low and high LET radiation, a wide range of solid tissue tumors develops, including those of the liver, lung and kidney. This range is in stark contrast to the tumor spectrum in the Trp53 null mice post-radiation, which mainly comprise lymphomas and sarcomas. The Trp53 deltaP mouse therefore represents an exquisitely radiation-sensitive model to study radiation-induced tumors similar to those seen in the human population.

We have previously identified a unique pattern of genomic alterations arising in low LET radiation-induced lymphomas in Trp53 null mice. In order to enable direct comparisons with our previous data we have successfully bred the Trp53 deltaP mice onto a 99.9% pure FVB/N background. We are investigating impact of high LET compared to low LET upon tumor latency and spectrum in mouse models with compromised Trp53 activity. We exposed 282 FVB/N mice of genotype homozygous Trp53 ?P, WT and Trp53 heterozygotes, to 50cGy of either 600 MeV/amu Fe particles or gamma radiation. Our results show a significant reduction in tumor latency in Trp53 delta P and Trp53 heterozygote mice exposed to high LET radiation, in comparison to mice of the same genotype exposed to low LET. Of note, there is a negligible impact of genotype (either Trp53 delta P or Trp53 heterozygotes) in tumor-free survival when mice are exposed to the same radiation quality. Unexpectedly, we have observed a high number of mammary tumors and skin carcinomas in the Trp53 heterozygote mice following exposure to high LET radiation. We have previously exposed Trp53 heterozygote mice on a range of different genetic backgrounds to various doses of gamma radiation and have not observed a high frequency of mammary or skin tumors. This suggests a radiation-quality specific shift in tumor spectrum. Further investigation of the potential mechanism by which exposure to high LET radiation causes this shift in spectrum and reduced latency is currently underway.

In order to gain a greater insight into the acute cellular response mechanism to exposure to high LET radiation we have performed a series of short time point experiments, with time points ranging between 6 hours and 1 week post 600MeV/amu Fe particle radiation. Preliminary analysis of microarray data from RNA of the dorsal skin suggests heightened expression of gene clusters enriched for acute inflammatory response at the 6 hour time point in the Trp53 delta P mice compared to the WT, this difference is not seen in the later time points. Conversely, genes associated with proliferation have much lower expression in the Trp53 delta P compared to WT at this 6 hour time point.

To investigate the possibility that the patterns of genetic changes in tumors induced by high and low LET radiation are different in Trp53 abrogated mice, we have begun an analysis of whole exome sequences of a range of tumors by radiation exposure. We have extracted DNA from 61 radiation-induced (either high or low LET) tumors of the mammary and skin, and thymic lymphomas in Trp53 delta P and Trp53+/- mice and sent these, along with normal control DNA samples, to our collaborator David Adams at the Sanger Institute, Cambridge, UK to carry out exome sequencing. Our data have identified candidate mutations in genes such as Notch1, Pten, Hras, Kras and Trp53, but further validation and confirmation of these mutations, as well as detailed analysis of the mutation spectrum, is in progress.

Project 3 uses agent-based models (ABM) to simulate the dynamic response of tissue to radiation in conjunction with Monte Carlo models of DNA damage to provide an integrated mechanistic model of the biology. This modeling approach is used to evaluate and prioritize key events during radiation-preceded carcinogenesis that result from immediate targeted DNA damage and persistent non-targeted effects.

As described in Project 1, a single low dose of radiation administered during puberty increases mammary stem cell frequency in mice at adulthood. We have developed modeling approaches to dissect the mechanistic basis of this observation. Our approach used ABM that integrates experimental data obtained from in vivo expression profiles and in vitro growth analysis of human mammary epithelial cells exposed to specific conditions. In silico modeling of mammary morphogenesis using ABM allows these events to be evaluated in the context of development, with its incumbent complexity of lineage heterogeneity, proliferation and morphogenesis.

It is not clear how processes induced by radiation affect stem cell regulation. Given the experimental challenges of testing whether radiation increases (1) cell inactivation (i.e. death) that then triggers proliferation; (2) signals for stem cell self-renewal, or (3) dedifferentiation, we turned to modeling to evaluate these potential mechanisms. Last year, we evaluated the contributions of radiation-induced cell death, senescence and self-renewal signaling. In the past year, we tested dedifferentiation and selective cell death as additional mechanisms for the increased mammary stem cell frequency in irradiated mice. We used these simulations to determine which mechanism when induced in a juvenile gland, had an impact on stem cell frequency measured at adulthood. The ABM rejected all modes of cell inactivation through cell death and replicative senescence as a possible means to increase stem cell frequency.

Tumor incidence in P53 compromised mice exposed to different radiation qualities is now being modeled. P53 deficiency is simulated by reducing cell-cycle checkpoint of epithelial agents, leading to increase genomic instability, increased mitotic catastrophe, and reduced apoptosis. Consequently, dividing epithelial agents which are genomically unstable have a probability of mutating a set of functional targets (essential housekeeping functions, non-essential functions, and functions that dictate cellular phenotype). In silico tumors (clones growing out-of-control in the model) emerge from these simulations and can be tracked back to MaSC agents that have lost anti-growth signal sensitivity and extracellular matrix regulation via successive mutations. When genomic-instability is combined with the transient increase in stem-cell self-renewal following IR, our ABM predicts a shorter latency for tumor incidence, as observed experimentally.

Second, we will develop a comprehensive data base of normal tissue and irradiated tissue by analyzing different organs in each mouse. This data base will become a public resource.

Third, our studies will identify genetic predisposition to radiation associated cancer, and may thus help protect humans who have the greatest risk of cancer.

This NSCOR provides experimental and modeling studies to define the efficiency and physiological context in which high LET radiation increases epithelial cancer risk. We hypothesize that both targeted and NTE contribute to radiation carcinogenesis. Targeted effects of ionizing radiation result from interaction of energy with DNA, leading to DNA damage responses and changes in genomic sequence. On the other hand, poorly understood NTE alter phenotype and multicellular interactions that could contribute to cancer. We hypothesize that the greater carcinogenic risk from high LET irradiation is due to NTE promotion of targeted radiation effects. The data generated the NYU/LBNL/UCSF NSCOR will provide a comprehensive analysis of LET dependence for solid tumor frequency in different tissues, integrated with computational modeling and systems biology to identify key events.

Project 1 Progress

We used radiation/genetic mammary chimera mouse model to separate targeted effects in the organ from the non-targeted effects on host cells. This model consists of unirradiated Tp53 null mammary epithelium transplanted into the inguinal cleared mammary fat pad of wild type host mice. Host mice were irradiated three days before transplantation with one of the following: 100 cGy gamma-rays, or 29 or 81cGy 600MeV/amu Fe corresponding to a fluence of approximately 1 and 3, particles respectively per cell, or 10, 29 or 81cGy 350MeV/amu Si corresponding to a fluence of approximately 1, 3 and 7 particles respectively per cell at the NSRL. Controls were sham-irradiated at NSRL. These mice were monitored for 600 days for subsequent development of palpable tumors. Upon detection of a first mammary gland tumor, a survival surgery was performed allowing the opportunity for a second tumor to develop on the contralateral gland. At experiment termination, we determined which transplants failed, which were eliminated from analysis.

To evaluate the contribution of the radiation effects on epithelium (so-called targeted effects like mutation), 204 fat-pads were cleared and transplanted with Trp53 null epithelium at 3 weeks of age were irradiated at 10 weeks of age at NSRL10-C with 30 cGy and 82 cGy 600Mev/amu Fe at or 10, 50, 100 cGy gamma-radiation. To evaluate the effect of age in the radiation chimera-experiment, mice were irradiated at 10 months, simulating the average age for an astronauts. For this purpose 200 mice were cleared at 3 weeks of age and then housed under standard conditions. At 10 months the mice were irradiated at NSRL 11B with three different ions (81 cGy Fe 600Mev/amu, 49 cGy Ar and 29 cGy Si 350Mev/amu) or three dose doses of gamma-radiation (10, 50, 100 cGy). Three days after irradiation the inguinal glands were transplanted with Trp53 null fragments. At this point the mice are being palpated for tumor detection and we expect to complete this experiment by fall 2012.

Project 2 Progress

To study the long-term impact of high LET compared to low LET radiation upon tumor formation we began a long-term tumorigenesis study in November 2010 using p53 deltaP mice, aged between 6-10 weeks, upon a mixed background (129/Sv/C57BL/6). 110 p53 deltaP mice were irradiated with either 29, 81, or 100 cGy 600 MeV/amu Fe-ion particles or 100 cGy gamma-radiation. In addition, wild-type controls of 129/Sv or C57BL/6 mice were irradiated. At the time of sacrifice radiosensitive tissues including the skin, mammary, thymus and spleen were harvested along with the arising tumors, which will be used for gene expression analysis, sequencing and histology. Furthermore, blood samples are taken for blood smears, complete blood count and pellets and serum. Our results show that over time there is no statistically significant difference between the 29, 81, 100 cGy Fe-particle exposure and 100 cGy gamma-radiation exposure upon tumor-free survival. The most prevalent tumor types seen were sarcomas, carcinomas and lymphomas. These have been seen in various organs including the skin, mammary gland, lung, liver, spleen, thymus and thyroid. At this point we have not observed a different pattern of tumor type or tissue of origin induced by different doses of Fe-ion or gamma-radiation. RNA and DNA are currently being extracted from these tumor samples for gene expression and sequencing analysis.

We have previously identified a unique pattern of genomic alterations arising in low LET radiation-induced lymphomas in Trp53 null mice. In order to enable direct comparisons with our previous data we have successfully bred the p53 deltaP mice onto a 99.9% pure FVB/N background. Currently, there are no studies describing the impact of high LET radiation upon the FVB genetic background, therefore these studies will provide valuable information as to the impact of genetic background upon heavy-ion radiation-induced carcinogenesis. In June and November 2011 we irradiated 250 FVB/N mice of genotypes p53 deltaP, and Trp53 wildtype and heterozygote mice with 50 cGy of either 600 MeV/amu Fe-ion particles or gamma radiation. These studies are still ongoing, and a range of tumors in radiosensitive tissues including the skin, liver, lung, mammary, thymus, thyroid and spleen has been observed in the p53 deltaP and Trp53 heterozygote mice. Currently we are extracting DNA and RNA from these tumors for gene expression and genomic analysis.

We completed a short-term time course in April 2012 using 125 mice. Consistent with previous experiments we used a dose of 50 cGy and irradiated with both 600 MeV/amu Fe-ions and gamma ions. Four genotypes of FVB/N mice were irradiated: p53 deltaP, Trp53 heterozygote, null and wildtype. These animals were then sacrificed at times 6, 24, 48 hours and 1 week post-radiation, along with an un-irradiated control set. At the time of necropsy a blood sample was taken and four tissues were harvested (skin, mammary, thymus and tail). Sections of these tissues were both fixed for histological analysis and frozen. RNA will be extracted from frozen tissues and gene expression levels at various time points post-radiation in the different genotypes will be investigated. This is an exploratory time-point experiment which will guide the time post-radiation at which we sacrifice our backcross population animals, for gene expression network analysis.

We proposed to perform a systems genetic analysis of susceptibility to tumors induced by high LET using mice from a backcross set up between Mus spretus and FVB/N p53 deltaP. Unfortunately, we have encountered unexpected practical difficulties with generating this p53 deltaP F1 backcross progeny. Breeding was problematic, with very few litters being born, and small numbers of pups within each litter. Furthermore, upon genotyping it became apparent that there was a significant level of lethality associated with the homozygous p53 deltaP backcross animals. This was surprising, as we have been able to breed the Trp53 null allele in a similar cross without seeing any evidence for lethality in this background. As a result, we were unable to breed sufficient numbers of F1bx Mus spretus/FVB/N p53 deltaP as would be required to have significant statistical power for analysis.

Hence we have changed the Spretus backcross studies to analysis of the Trp53 heterozygous mice. These data will also directly complement the studies in Project 1 which use a Trp53 null mouse model. Currently we are initiating a backcross between Mus spretus and FVB/N Trp53 heterozygotes, which we will expose to 50 cGy 600 MeV/amu Fe-ions and ion-radiation. From the tissue samples taken pre- and post-radiation we will extract RNA and investigate the perturbations in tissue gene expression networks after exposure to high and low LET radiation. This will allow us to generate a unique view of gene expression networks and enable us to identify components that impact upon sensitivity to tumor development post-exposure to both low and high LET radiation. Ultimately, these data, together with the genomic analysis of tumors induced by both high and low LET radiation described above, will assist in the identification of potential susceptibility genes and pathways for future assessment in human populations.

Project 3 Progress

As described in Nguyen et al., a single low dose of radiation increases mammary stem cell frequency in mice. It is not clear how processes induced by radiation affect stem cell regulation. In order to evaluate the long-term impact of radiation-induced death/senescence and stem cell self-renewal signaling in the mammary gland, so we integrated agent based modeling (ABM) with in vivo measurements and in vitro cell culture data. In silico modeling of mammary morphogenesis using ABM allows events to be evaluated in the context of development, with its incumbent complexity of lineage heterogeneity, proliferation and morphogenesis.

Expression profile data from Project 1 were analyzed from mammary glands of mice irradiated 1, 4 and 12 weeks before by sparsely ionizing gamma-ray radiation (1 Gy) or densely ionizing 350 MeV/amu Si particle or 600 MeV/amu Fe particle. We identified enrichment for a MaSC signature at 1 week post-IR regardless of radiation quality. At 4 weeks post-IR, the number of MaSC genes decreased. At 12 weeks, MaSC gene enrichment was no longer evident in any group. These analyses are ongoing in collaboration with Project 1.

We benchmarked the ABM using mammary gland measurements (e.g. branching frequency, time of development). Model parameters (e.g. cell doubling time, differentiation probabilities) were determined by generating simulations matching in vivo measurements from tissues representing maturation from puberty to adult (weeks 3 to 9). We then tested various mechanisms thought to be involved in the radiation-induced stem cell enrichment. ABM rejected cell loss via death or replicative senescence as a source for stem cells increase, and supported a small, transient increase in TGFß and Notch activity as an effective means to stimulate self-renewal.

Preliminary analysis of expression profiles of tumors from Project 1 suggests that many biological processes were affected by HZE host irradiation. Pathway analysis of the genes associated with tumors from Si-irradiated host invoked cell growth and proliferation, hematological disease, hematological system development, cellular assembly and organization, and metabolic disease. These tumors were also significantly enriched for a mammary stem cell signature. These processes and signatures are similar to those we previously identified in tumors arising in gamma-irradiated hosts (Nguyen et al. 2011. Radiation Acts on the Microenvironment to Affect Breast Carcinogenesis by Distinct Mechanisms that Decrease Breast Cancer Latency and Affect Tumor Type. Cancer cell, In press).

Bioinformatics Components

We switched from Affymetrix Mouse 430 2.0 arrays to Affymetrix Mouse gene 1.0 ST arrays at the UCSF Cancer Center. The latter is about half the cost, due to its ability to process more samples simultaneously. This newer technology is also fully compatible with the older arrays and in fact brings new elements that will benefit this proposal. However, there are fewer analytic tools for this platform so we have begun to generate our own analytical tools for the information from Mouse Gene 1.0 ST Arrays.

Second, we will develop a comprehensive data base of normal tissue and irradiated tissue by analyzing different organs in each mouse. This data base will become a public resource.

Third, our studies will identify genetic predisposition to radiation associated cancer, and may thus help protect humans who have the greatest risk of cancer.

The second year of the funding was devoted in part to staffing and training the three projects encompassed in this NASA Specialized Center of Research. An additional postdoctoral fellow, Cassandra Adams, was recruited to work on Project 2, and has completed training for BNL and NSRL access and use. The PI and co-investigators met three times to discuss the research.

In the second year, we have completed three experiments at NSRL using three particle beams. We have irradiated.

Project 1

We use a novel radiation/genetic mammary chimera in which either the whole mouse, the host microenvironment, or the transplanted epithelium is irradiated. Preliminary data show that NTE dramatically increases tumor development in hosts irradiated prior to being transplanted with p53 null mammary epithelium. This model will be used to evaluate the relative contribution of targeted vs non-targeted radiation effects from low vs high LET and to determine whether the contribution of NTE is impacted by age, or that chronic low dose rate exposure to low LET (i.e. protons) will interact with acute high LET radiation. Aim 1 consists of tumor incidence and latency measured for gamma-rays and two fluences of three ions (600 MeV Fe and 350 MeV Si or Ar) that will require two years to complete for each ion. Two experiments consisting of hosts irradiated with either Fe or Si particles have been initiated during the Fall 09 and Spring 10 runs.

The data from the radiation chimera model on which Project 1 is based was published in the May 2011 issue of Cancer Cell, which co-authored by NSCOR members, Drs. I. Illa-Bochaca, Mao and Barcellos-Hoff. We report that mice irradiated with low doses (10-100 cGy) and subsequently orthotopically transplanted with p53 null mammary epithelium undergo accelerated tumorigenesis of predominantly estrogen receptor negative tumors. One of the most notable new findings resulting from this study is that radiation-induced activation of Notch signaling may contribute to altered stem cell self-renewal. Preliminary data suggest that the size of the stem/progenitor compartment is also affected by densely ionizing radiation. Notch and beta-catenin immunolocalization in particle irradiated tissue is being conducted to determine how radiation quality affects the magnitude of this effect.

The contribution of inflammation as a mediator of NTE carcinogenic risk studied by Drs. Demaria and Matsumura in Aim 2 using systemic and local functional assays. Overall, data suggest that all radiation regimens cause a mild inflammatory response. The number of total spleen cells was detected in mice treated with gamma or high dose Fe at 1 week was reduced by 50%, with return towards the sham level at 1 month. Microarray expression profiling resonates with the immune function studies in that inflammation associated genes are enriched in the expression profile of mammary gland post-irradiation.

Project 2

The goal of this project is to use genetic analysis as an unbiased means to reveal the contribution of different biological components and/or processes. We hypothesize that organ-specific risks have different genetic determinants/associations, and that radiation quality differentially affects the genetic architecture of gene expression in radiosensitive tissues. Expression profiles in tissues that are sensitive to the effects of radiation generate a network view of gene expression. We will investigate how this network is perturbed by genetic manipulation (alterations in the p53 pathway), and by exposure to high or low LET radiation. To facilitate comparisons with earlier gene network analysis, the p53 delta-P mouse is currently being bred onto a pure FVB background for further timepoint and longer term tumorigenesis studies, which commenced in 2011. In addition, we will investigate responses of p53 knockout mice on various genetic backgrounds to high and low LET radiation.

Project 3

To provide an integrated mechanistic model of the biology, Monte Carlo models of DNA damage will be used with agent-based models (ABM) to simulate the dynamic response of tissue to radiation. This modeling approach is used to evaluate and prioritize key events during radiation-preceded carcinogenesis that result from immediate targeted DNA damage and persistent non-targeted effects. Extended time series will use gene expression motifs identified for radiosensitive tissues in Projects 1 and 2 translated into functions (proliferation, apoptosis) of either the agents or the microenvironment. we have constructed a computational systems model of the mouse mammary gland using an agent-based formalism. Agents are autonomous software objects that have defined rules or mathematical equations that determine how they behave and interact with each other and their local environment. Our in silico mammary gland consists of interacting agents simulating either mammary epithelial cells or stromal cells, in a simulation environment with variables representing extracellular matrix properties. We have modeled normal murine mammary morphogenesis and maintenance by establishing a set of rules and parameters that led to cumulative agent behavior that match experimental data of the developing gland. Our next objective is to model how radiation can perturb each of mammary gland compartments and lead to a disorganized morphogenesis, i.e. dysplasia, a hallmark of cancer.

In combination with modeling microdosimetry of various radiation qualities in this system, this tool will allow us to test how geometrical properties of energy deposition can modify relative biological effectiveness. This exercise will help refine hypotheses in which radiation carcinogenesis is promoted by synergistic interaction between targeted damage and non-targeted effects.

In the first year, we have completed two campaigns at NSRL using two particle beams consisting of approximately 300 mice. Six organs were collected at three or more time points for analysis. Mammary gland tumors will be monitored for the next year in mice receiving p53 null epithelial transplants after irradiation and compared to controls irradiated with low LETat BNL.

This study is significant because the mechanistic analysis of biological and genetic determinants will improve understanding of radiation quality effects. Aside from defining how radiation affects the very processes by which carcinogenesis advances, the information gained should provide new, interaction-based avenues for mechanism based countermeasures for space radiation exposure.

This NSCOR focuses on experimental and modeling studies to define the efficiency and physiological context in which high LET radiation increases epithelial cancer risk. Experimental data suggest that radiation carcinogenesis is a two-compartment problem: ionizing radiation can alter genomic sequence as a result of damage due to targeted effects from the interaction of energy and DNA, and alter phenotype and multicellular interactions that contribute to cancer by poorly understood NTE. This study hypothesizes that radiation NTE create the critical context to promote cancer and that inflammation is a key process, and that low vs. high LET cancer susceptibility will depend on specific genetic backgrounds. Defining the relative contributions of targeted vs. NTE to carcinogenesis and the genetic determinants of the high LET cancer susceptibility will improve understanding of radiation carcinogenesis and will help reduce uncertainties in NASA radiation risk assessment.

The study is divided into three Projects, each of which is described in detail in the original proposal. The projects listed in the proposal are as follows:

• Project 1 will test use novel transgenic and humanized mouse mammary models to evaluate the relative contribution of radiation targeted and NTE to carcinogenesis as a function of radiation quality and of age a radiation chimera model. Inflammation will be monitored as a key process affected by radiation NTE and conduct a systems biology analysis of mammary gland.

• Project 2 will employ new systems genetic analysis of cancer susceptibility in radiation sensitive organs to define critical regulatory networks that impact the relative carcinogenic effectiveness of high LET radiation.

• Project 3 will use Monte Carlo simulations of targeted radiation effects and agent based modeling of cell interactions to model the relative contributions of targeted and NTE to carcinogenesis.

This NSCOR will provide important data on radiation quality dependence of targeted and non-targeted radiation effects and the genetic determinants of cancer susceptibility, which together will be used to reduce uncertainties in NASA radiation risk assessment.