Skip to main content

Studies on a cohort of Serbian children exposed to x-irradiation to determine the contribution of the non-coding genome to
susceptibility at low doses

Final Report Summary - DARK.RISK (Studies on a cohort of Serbian children exposed to x-irradiation to determine the contribution of the non-coding genome to susceptibility at low doses)

Executive Summary:
Estimates of risk at low doses of radiation can be improved by studying individuals exposed at an early age, where there are no underlying medical conditions and where biological material is available for molecular epidemiological studies. We have established the Serbian Tinea Capitis Cohort, composed of more than 25 000 individuals exposed as children to X-irradiation of the head. In an on-going effort we have created a digital registry of over 9000 individuals linking exposure data with government records and medical databases. Of these more than 1000 individuals have agreed to be interviewed and to provide biological material and grant access to their medical records. Standard operating protocols have been created and validated for the collection, storage and distribution of biological samples. These materials have proven acceptable for the analysis of non-coding RNA levels in the circulation, opening the way for biomarker identification. In parallel studies we have established that both long non-coding and micro RNAs are regulated by radiation exposure, and that they influence a number of regulatory pathways, including metabolism, epigenetic markers, telomeric integrity and cell division. These studies provide a new set of candidate biomarkers for identifying radiation responses.

Dark.Risk has delivered the following key results:

- Established the Serbian Tinea Capitis Cohort of 25 000 individuals receiving low dose radiation exposures.
- A validated SOP for sampling, storing and transporting biomaterials for radiation biomarker analysis.
- Mapping of the non-coding transcriptome responses to 20mGy and 100mGy low dose radiation exposures.
- A set of novel non-coding RNAs have been identified as candidate radiation biomarkers.

Project Context and Objectives:
The dose limits for exposures to ionizing radiation are based upon internationally accepted estimates of the risk of adverse health effects. The scientific evidence underpinning dose limits continually evolves, but can only provoke legislative responses when sufficiently robust, in particular when validated by epidemiological evidence. An area where MELODI has recognised an urgent need for deeper understanding concerns the contribution of individual differences in sensitivity. The response of an individual to radiation exposure is neither constant over a lifetime, nor uniform between individuals. However, the extent of the influence of the individual genetic constitution is uncertain. MELODI has determined that a combination of molecular epidemiology and improved mechanistic understanding offers the best opportunity to quantify the contribution to risk. Three roadblocks hinder realization of the MELODI objectives. One comes from the lack of epidemiological cohort(s) that combine comprehensive health information, reliable dosimetry and the availability of biological materials for analysis. The second is the lack of even a basic understanding of how individual differences in sensitivity are genetically determined. The third is a lack of suitable biological markers characterising individual radiation exposure and radiation response. Dark.Risk has set as its goal the removal of all three of these roadblocks.

Objective 1:
The Serbian Registry of Tinea Capitis Children (SRTCC) - A novel epidemiological cohort for European low dose research
Tinea Capitis (ringworm of the scalp) is a fungal infection of the hair roots, most prevalent among school-age children. Prior to introduction of Griseofulvin in 1960 various methods were used to facilitate removal of the hair to allow topical application of remedies. The Adamson-Kienbock protocol used X-irradiation of the scalp for epilation, and was a common treatment in the first decade of the 20th century (1). The treatment protocol was highly standardized, with five overlapping exposures being administered over 5 days. The treatment used unfiltered 75-100 kVp X-rays such that children were typically exposed to a 3.5–4Gy scalp dose. Dosimetric studies in Israel using an original X-ray machine and a head phantom of the skull of a 7 year old have estimated the average dose to the brain at 1.5 Gy (2), to other parts of the upper body considerably less. The first retrospective cohort study of over 11,000 Israeli subjects demonstrated increased risk both for benign and malignant head and neck tumours. Other reports followed, increasing the list of radiation-induced tumours to include head and neck, skin and breast (3-7).

In the aftermath of World War 2 the incidence of Tinea Capitis reached epidemic proportions in many countries, including the Former Yugoslavia. Between 1950-1959 a UNICEF-sponsored programme treated almost 100,000 children using X-irradiation. In 2005, collaboration between Serbian and the Israeli scientists reviewed the programme (8). This resulted in the securing of 24,000 case records from the hospital registry of the Mycosis Hospital in Belgrade. ROSA has conducted a feasibility study of the available information. This has revealed that it is possible to identify individuals from public records, and that cancer registry data can be used to identify cases within the sample. Thus, we have a realistic chance to ascertain not only health outcomes and applied dose, but also to obtain biological materials from living individuals. These materials, as well as the epidemiological data (causes of death, confounding factors, biometric parameters etc.) will be used to create a new epidemiological cohort - The Serbian Registry of Tinea Capitis Children (SRTCC). This resource will provide a focus for European researchers for many years to come, as it will allow molecular epidemiology studies on a defined cohort. In Dark.Risk our objective has been to establish the low dose epidemiological cohort of individuals exposed as children to cranial x-irradiation in Serbia (Serbian Registry of Tinea Capitis Children (SRTCC)). With this cohort we will conduct a feasibility study to establish if the cohort is suitable for the collection of biological samples for molecular epidemiological studies. This will ultimately lead to the creation of a biobank containing samples from the irradiated individuals and appropriate control individuals.

Objective 2:
Deciphering the new biology of the non-coding genome as a means to understand individual sensitivity to radiation
The effort to identify genes responsible for modifying an individual's risk of developing sporadic cancer has been highly successful in finding the genes responsible for monogenetic traits (familial cancer syndromes). In contrast, though, searches for the genes making a more complex genetic contribution (e.g. low penetrant modifying loci) have almost universally failed. Genome-wide analysis studies (GWAS) have consistently placed the loci governing susceptibility to cancer into regions of the genome lacking translated (protein coding) mRNAs. Only in the last two years has it been recognised that non-coding RNA transcripts may be responsible for these genetic effects. Less than 5% of the human genome is transcribed into protein coding messenger RNA. The remaining 95% is non-coding and, apart from the ribosomal and transfer RNAs, was long considered to be inert and therefore biologically irrelevant. Only with the advent of advanced deep sequencing has it become clear that almost all of this non-coding genome is in fact actively transcribed into non-coding RNAs and that these have the ability to influence cellular function. This has led to the naming of the non-coding transcriptome as the “dark matter" of the human genome. A number of different functional classes of non-coding RNA transcripts are recognised, two classes of which (microRNAs and long non-coding RNAs) show involvement in radiation responses. (9). The stability of mRNAs, and hence the length of time an RNA is available for translation into protein, is regulated by a complex family of over 1000 short length RNAs, the microRNAs (10). Each microRNA is itself transcriptionally regulated and is capable of targeting and removing multiple mRNA species, leading to highly coordinated rapid changes in cellular phenotype. Our studies have revealed that microRNAs are highly regulated by exposure to radiation, and that they are essential for the survival of irradiated cells (11).

Long (intergenic) non-coding RNAs (lincRNAs) are spliced, polyadenylated gene transcripts that do not encode protein. Their action appears critical for chromatin structure, in particular transcriptional repression and telomeric structure. One of the first lincRNAS described was the metastasis associated HOTAIR-lincRNA found to be highly up regulated in breast tumours. Although, the precise mechanism of HOTAIR activities remains to be elucidated, it appears that this lincRNA acts as a molecular scaffold for guiding histone modification complexes to specific genomic targets (12). A role for the lincRNAs in radiation responses has been suggested following the demonstration by Huarte et al (13) that lincRNA-p21-is responsible for switching off the p53-mediated transcriptional response to DNA damage. lincRNA-p21 is located 15kb upstream of CDKN1A (p21) gene. At the CDKN1A promoter, five lincRNAs, similar to the CDKN1A mRNA itself, are induced by DNA damage. One of these lincRNAs, named PANDA, is a non-spliced lincRNA that is transcribed antisense to CDKN1A and during DNA damage shows also p53 dependent response (14). The non-coding TERRA lincRNAs transcribed from the sub-telomeric regions of the genome have also been implicated in the radiation response. We have shown that the susceptibility modifying Rb1 gene sustains telomere maintenance by transcribing TERRA. Impairment of Rb1 function leads to increased sensitivity to radiation carcinogenesis and is mediated by changes in TERRA expression and histone methylation on the telomeres that lead to truncated telomeres and genomic instability (15). In Dark.Risk we will experimentally determine if the "dark matter" contributes to the long-term health effects of ionizing radiation at doses below 0.1 Gy. This will be done through biological experiments to establish function, and through the identification of non-coding genome biomarkers of exposure and response. Variations in the expression of non-coding RNAs will be examined to determine if there is evidence for a role in individual sensitivity for non-coding RNAs other than TERRA.

Objective 3:
New molecular markers indicative of radiation exposure and of radiation response
MicroRNAs are released into the circulation within microscopic vesicles (exosomes). Their function is unknown, but initial indications suggest they may be reliable indicators of tumour growth and tumour response to treatment. Indeed, we have observed changes in the level of microRNAs present in the circulation in a small sample of radiotherapy patients (16). In Dark.Risk we will examine the potential of the non-coding RNAs to deliver a new generation of biological markers for use in molecular epidemiological studies.

1. Adamson H. A simplified method of x-ray application for the cure of ringworm of the scalp: Keinbock's method. Lancet 1909;1:1378-1380.
2. Werner A, Modan B, Davidoff D. Doses to brain, skull and thyroid, following x-ray therapy for Tinea Capitis. Phys Med Biol 1968;13:247-258.
3. Shore RE, Moseson M, Xue X, et al. Skin cancer after X-ray treatment for scalp ringworm. Radiat Res 2002;157:410-418.
4. Modan B, Baidatz D, Mart H, et al. Radiation-induced head and neck tumours. Lancet 1974;1:277- 279.
5. Albert RE, Omran AR, Brauer EW, et al. Follow-up study of patients treated by x-ray for tinea Capitis. Am J Public Health Nations Health 1966;56:2114-2120.
6. Munk J, Peyser E, Gruszkiewicz J. Radiation induced intracranial meningiomas. Clin Radiol 1969;20:90-94.
7. Modan B, Chetrit A, Alfandary E, et al. Increased risk of breast cancer after low-dose irradiation. Lancet 1989;1:629-631.
8. Shvarts S, Sevo G, Tasic M, et al. The tinea Capitis campaign in Serbia in the 1950s. Lancet Infect Dis 2010;10:571-576.
9. Esteller M. Non-coding RNAs in human disease. Nat Rev Genet 2011;12:861-874.
10. Kasinski AL, Slack FJ. Epigenetics and genetics. MicroRNAs en route to the clinic: progress in validating and targeting microRNAs for cancer therapy. Nat Rev Cancer 2011;11:849-864.
11. Kraemer A, Anastasov N, Angermeier M, et al. MicroRNA-mediated processes are essential for the cellular radiation response. Radiat Res 2011;176:575-586.
12. Gibb EA, Brown CJ, Lam WL. The functional role of long non-coding RNA in human carcinomas. Mol Cancer 2011;10:38.
13. Huarte M, Guttman M, Feldser D, et al. A large intergenic noncoding RNA induced by p53 mediates global gene repression in the p53 response. Cell 2010;142:409-419.
14. Huang T., Wang Y., Lin M., et al. Extensive and coordinated transcription of noncoding RNAs within cell-cycle promoters. Nature Genetics 2011;43(7):621-9.
15. Gonzalez-Vasconcellos I., Anastasov N., Sanli- Bonazzi B., et al. Rb1 haploinsufficiency causes telomere attrition and radiation-induced genomic instability. Oncogene 2012;Submitted.
16. Niyazi M, Zehentmayr F, Niemoller OM, et al. MiRNA expression patterns predict survival in glioblastoma. Radiat Oncol 2011;6:153.

Project Results:
Scientific results of the Dark.Risk project: Development of a new epidemiological cohort for the study of late effects of low radiation doses: The Serbian Registry of Tinea Capitis Children

Epidemiological studies have focused on cohorts of individuals exposed to a number of radiation exposure scenarios, from nuclear workers to survivors of the atomic bombings. Each cohort offers benefits and advantages, with uncertainties in both dose and outcome. Studies of patients exposed to therapeutic medical irradiation have provided valuable evidence of the risks of radiation, in particular of non-cancer effects. The cohorts of patients that have been used in these studies are, however, of limited use in evaluating the risks at low doses, as they typically receive high doses in conjunction with adjuvant therapies for existing medical conditions.
In the late 1950s public health initiatives in a number of countries mandated x-ray treatment for the depilation of the scalp as the standard of care for the treatment for children infected with the Tinea Capitis fungal disease of the hair roots. This internationally standardized treatment protocol proved highly effective in combating the fungus. Only decades later was it noted, firstly in Israel, that there was an increased frequency of a number of cancers in the treated individuals. The standardized treatment and narrow age-distribution at exposure indicate that a study of health effects in Tinea Capitis patients would be a powerful epidemiological approach to study risk of low dose irradiation. A team of epidemiologists led by Goran Sevo and Marija Tasic in Belgrade previously uncovered documentation recording the X-irradiation treatments made to over 25000 Serbian children. If the exposed individuals were traceable and willing to provide biological samples, the potential for long-term follow-up with biomarker analysis offers hitherto unprecedented opportunities for low dose research.
Work package 1 was intended to establish The Serbian Registry of Tinea Capitis Children by organising the transfer of written documentation from hospital registers into a digital archive. Individual patient records were to be established using medical files, local citizen registration and in the case of living patients, individual interviews to establish identity, cofounders and the willingness to contribute biological samples. In the second stage a database was to be prepared to link to the archive to the Serbian Cancer Registry. Finally to establish the usefulness of the cohort for molecular epidemiology a subset of the cases were to be selected, and the individuals contacted to validate clinical data and, if possible, to provide blood samples for the analysis of non-coding RNA. This latter set of tasks was designed to determine which, if any, problems in logistics, sampling, consent and analytical technology exist.

Establishing the registry.
The essential ethical approvals were obtained at different stages of this work from the Serbian Medical Society, Serbian Institute of Health and the Institute of Gerontology and Palliative Care, Belgrade. A letter of support for the essential identification activities was obtained from the Serbian Ministry of Health, inviting local health care facilities to take part in the cohort enrollment. We based our entitlement to privacy data relevant in different stages of this work on the following legal framework: (1) Freedom of Information Act, and (2) on articles 6, 12 and 14 of the Law on Protection of Personal Data ("Službeni glasnik RS", br. 97/2008, 104/2009, 68/2012 - Odluka Ustavnog suda).
Enrollment is currently recorded into a separate data set to be linked with the existing SRTCC database via Unique Study Number (“primary key”), allocated to each BCMHF record and maintained throughout the study. Due to a recent policy change in the protection of privacy ROSA was officially advised by the Commissioner for Information of Public Importance and Personal Data Protection (Serbia) that: (1) no further activity was to be carried out on patients' identity check within the current framework, and (2) different approaches to the initially proposed ones must be used in selecting controls from the general population and from the unexposed siblings of radiation exposed individuals. This change must be subject to a renewed ethical clearance. The requested renewed ethics approval has been obtained, and follow up to address other Commissioner's recommendations is currently under way.

Current status of the epidemiological study cohort.
Out of 25 166 patients’ records at the Belgrade Children's Mycosis Hospital from 1950-1960 we have to date confirmed the identity of 13868 individuals from public and municipal sources (LROs), making these persons eligible to cohort enrolment. Cohort recruitment was initiated by tracing individuals aged 0-15 at the time of exposure. The medical staff of community health centres (primary care physicians and community nurses) carried it out locally. At the end of the Dark.Risk project 1196 irradiated individuals have been fully enrolled (approached, interviewed and with individual approval for the study to access to their medical records).
Linking patient details to the Serbian Cancer Registry has been completed for all of the identified individuals. Successful matching was obtained only for 9435 persons whose JMBG records were available. Due to unresolved transliteration format of the central registry no successful matching was possible using text variables (name, surname, place of birth etc.). Of the 9435 established records 1101 had been enrolled in the Serbian Cancer Registry database by 2015. The causes of death for 3498 deceased individuals are currently being evaluated against local registry offices' reports, the Serbian Cancer Registry, and the community health centres' reports. Amongst identified patients, 5628 siblings received irradiation for Tinea Capitis. Once data collection is completed this will provide an opportunity for nested risk analysis.

Identification of a sub-cohort of multiply exposed individuals.
Significantly, amongst the identified subjects there were 329 individuals who had undergone two or more irradiation treatments. Provided their willingness to be enrolled in the cohort they will provide a valuable sub-sample for dose-response analysis.

Feasibility study for establishing a biological repository.
A feasibility study has been successfully completed by approaching 514 identified individuals (approached, interviewed and obtained access to their medical records), and collecting biological samples from a randomly selected group of 31, in addition to collecting samples from 20 general controls (unexposed). In complimentary studies described below (work package 3) a set of Standard Operating Protocols were developed for collection, storage and transport of these samples. Biological samples of venous blood and urine were obtained, transported to the Institute of molecular genetics and genetic engineering, Belgrade (IMGGE), processed and stored at -80°C. Small aliquots of these samples were subsequently evaluated for their suitability for molecular analysis.
As a result of the project the Serbian Registry of Tinea Capitis Children (SRTCC) now contains the following three data sets:
(1) Belgrade Children's Mycosis Hospital File (BCMHF) – hospital registry transcribed into electronic format and evaluated for duplicated entries, sibling status and hospitalization details.
(2) Identification File – personal details of all patients whose identity was confirmed by local and municipal sources.
(3) Dates File - where all chronological variables from (1) and (2) were evaluated on an individual case basis to enable calculating different time variables (age on admission, length of hospitalization, etc.).

A new era of molecular epidemiology requires biological markers -The dark matter of the genome emerges into the light.
In theory, molecular epidemiology offers the possibility of focusing on the cases arising in any cohort that are actually caused by radiation, instead of loosing this information amongst the preponderance of sporadic (non-radiation caused) cases. This would confer the missing statistical power needed to evaluate low dose risk. A series of biomarkers would be needed that distinguish between radiation-associated and radiation-independent cases (i.e. biomarkers of causality). In many situations it is also known which individuals received radiation exposure, but the exact dose is often unclear, especially if considerable time has passed since the event, or if dosimetric data is missing. Molecular markers indicating past exposure would be an ideal way to improve the uncertainties in this area (i.e. biomarkers of exposure).
The past five years have seen an explosion in knowledge of the structure and function of the human genome, forcing a paradigm shift in the way we look at the information encoded in our DNA. Previously, 90% of the genome was thought to be simply spacer, needed only to ensure the correct functioning of the remaining 10% where the genes were positioned. It now turns out that most of the spacer (~90%) is actively transcribed into RNA in a regulated manner. The lack of a known function for this RNA prompted it to be termed the “dark matter” of the genome.
The level of metabolic energy required to produce the RNA, as well as evidence for evolutionary conservation, both suggest that there is probably a strong biological reason for the existence of the dark matter, now more correctly termed non-coding RNA. A number of recent studies suggest that the expression of the non-coding RNA is highly regulated during responses to radiation, and that distinct functions can be ascribed to the different components of the non-coding transcriptome. In particular the non-coding microRNAs appear to be highly stable in biological materials. A recent relevant observation is that the pattern of expression of non-coding microRNA can be used to both predict disease appearance and outcome.
Although, the data of the radiation effects on non-coding RNAs and deeper epigenetic control are in their infancy, recent studies from our and other laboratories confirmed the importance of the non-coding RNA modulation in the cells after irradiation and their potential to be used as efficient biomarkers in the future. In particular non-coding microRNAs appear to be highly stable in biological materials (ex. blood samples, buccal swabs, hair follicles etc.). A whole new field of serum RNA transcriptomics has opened up in the past year, delivering a range of disease markers based on non-coding RNA. In Dark.Risk a series of experimental models were used to establish the extent to which the non-coding transcriptome is involved in the radiation response, and how such a response can influence our understanding of the linear (or non-linear) dose response relationship. We designed the project to address the hypothesis that intracellular and extracellular changes in the non-coding RNA profile respond to irradiation at low doses. From this we tested the hypothesis that an LNT relationship existed.

Mapping the response of the non-coding RNA transcriptome to radiation.
Recent high throughput transcriptome analyses have identified a substantial portion of transcripts that are non-coding RNAs (ncRNAs) in mammalian genomes (Rinn and Chang, 2012). The ncRNA class contains two major groups, the micro RNAs (miRNAs) and the long non-coding RNAs (lncRNAs). Both groups are emerging as new classes of regulatory players, with involvement in diverse cellular functions. However, their underlying mechanisms remain poorly understood. As both potentially target protein levels these regulatory molecules may explain the frequently observed dichotomy between transcriptional activity measured by microarray and protein expression measured by mass spectrometry. The microRNAs have already been shown to play a role in regulating cell responses to radiation, the lncRNAs are untested.
The contribution of micro RNAs (miRNAs) to the radiation response was screened using Low Density Arrays and MCF10A cells as immortalized but not transformed human mammary epithelial cells. Figure 1 shows that most of the miRNAs measured were down-regulated 4 hours after low or medium dose irradiation, whereas at the 24hour time point, most of the detected changes in miRNA levels show a slight increase after irradiation.
From this initial array screening six miRNAs (hsa-miR-21, hsa-miR-221, hsa-miR-222, hsa-miR-125a, hsa-miR-494 and hsa-miR-335) were selected for validation of miRNA expression. All six miRNAs were down regulated 4 hours after irradiation. At the 24hour time point, the hsa-miR-21 and hsa-miR-494 levels were returned to normal, whilst the other four slightly overshot the resting levels.
In order to confirm how miR-21 expression was changed upon ionizing radiation, expression levels were quantified by qRT-PCR, 4 and 24 hours after low (0.25 Gy) or medium (2.5 Gy) dose of radiation using MCF-10A epithelial-like and MDA-MB-361 breast cancer cells. Interestingly, the expression profile between the two cell lines was changed in different ways. While in MCF-10A miR-21 levels were significantly up regulated after low and medium doses of radiation, in MDA-MB-361 cells miR-21 was either down regulated or not changed after low or medium doses of radiation. In order to down-regulate miR-21, a lentiviral approach using anti-miR-21 was applied. Combined miR-21 inhibition and radiation effects were analysed by exploring cell proliferation efficiency (Figure 2).
In MCF-10A cells significant increase in proliferation was detected 48 hours after irradiation (0.25 Gy or 2.5 Gy), while the proliferation capacity was decreased in MDA-MB-361 cancer cells after combined anti-miR-21 and radiation treatment. The differences in cellular response to anti-miR-21 treatment in combination with radiation were confirmed by using clonogenic survival and 3D microtissue growth delay assay (Anastasov et al., BMC Cancer, 2015).

Characterisation of long noncoding RNA (lncRNA) expression - PARTICLE
To identify key lncRNAs responding to low dose gamma irradiation, ArrayStar expression profiling was undertaken using three human cell lines (HUVEC, U2OS and T47D). Differentially expressed lncRNAs (87 up- and down-regulated) were identified from all three cell lines, through fold change filtering and irradiated versus non-irradiated comparative analysis between samples. A subset of 12 lncRNAs associated with genes involved in cancer, were selected for verification (Figure 3).
The focus of this part of Dark.Risk project has been on deciphering the mechanistic basis of one lncRNA recently designated the name PARTICLE (Hugo Gene Nomenclature PARTICL - 'Promoter of MAT2A-Antisense RadiaTion Induced Circulating LncRNA': NCBI reference sequence NR_038942.1) and results were published during 2015 (O’Leary et al., Cell Reports, 2015) with corresponding press release ( latest-news/press-information-news/article/26741/index.html). PARTICLE is located in an antisense orientation to the methionine adenosyltransferase 2A (MAT2A) promoter. MAT2A encodes the catalytic subunit of methionine adenosyltrasferase (MAT) – the enzyme responsible for the production of s-adenosylmethionine (SAM), the principal methyl donor used for DNA methylation (Mato et al., 1997). The orchestrated interplay between PARTICLE and MAT2A has been unravelled, involving functional responses to irradiation that are dependent on sub-cellular context. In-situ hybridisation utilising fluorescently labelled probes complementary to PARTICLE in MDA-MB-361 cells confirmed that low dose irradiation induce PARTICLE expression (Figure 3). It was found that PARTICLE and MAT2A transcripts co-localize as early as 4 hr. after low dose irradiation.
Furthermore successful PARTICLE knockdown using lentiviral knock-down silencing approach has been established within the project. Knockdown of PARTICLE caused over-expression of MAT2A in non-irradiated MDA-MB-361 cells (Figure 4). The increase in MAT2A expression was far greater following low (0.25 Gy) dose irradiation in comparison to medium (2.5 Gy) dose exposure (5.2 fold higher, p < 0.005). These findings highlight the various transcriptional response profiles emanating from exposure to either low or medium irradiation dosage and emphasize the homeostatic repressive influence of PARTICLE on MAT2A expression. The MAT2A promoter contains a CpG island differing in methylation levels followed by the irradiation exposure. The low (0.25 Gy) and medium (2.5 Gy) dose correlation between methylation and PARTICLE expression suggests the lncRNA may be influencing the methylation status of the MAT2A CpG island. Triplex-forming oligonucleotides within PARTICLE were predicted to bind upstream of a CpG island that is located in the promoter of MAT2A. Therefore, recently published data strongly support the triplex mechanism in the interaction of PARTICLE with the MAT2A DNA region. It is hypothesized that the formation of the triple helix (triplex) may serve as a potential ‘riboswitch’ for the binding of proteins or metabolites with PARTICLE serving as the central component of a dynamic RNA regulatory element responding to irradiation.

Radiation-induced expression of long non-coding RNA in breast primary epithelial cells.
The contribution of lncRNAs to radiation-induced damage response was screened using SurePrint G3 Human Gene Expression v2 Microarray (Agilent) on BPECs. Figure 5 shows that most of the lncRNAs were down regulated 2 hours after exposure to low or medium doses of radiation.
Irradiation of BPECs at 2 and 0.1 Gy induced changes in the expression of a common set of lncRNAs, which was not shared with the samples irradiated with 0.02 Gy. Specifically, whilst 17 of the 21 lncRNA genes regulated after exposure to 0.1 Gy were also regulated after exposure to 2 Gy, none of the 36 genes regulated at 0.02 Gy changed their expression in the 0.1 Gy sample, and only 2 in the 2 Gy sample. Similar radiation-associated trends were observed in the transcription profiles of protein-coding genes, which according to gene ontology analysis had a significant enrichment in pathways relevant for DNA damage response after the 2 Gy dose. Altogether, significant differences are noted in transcriptional-response profiles yielded by low doses of radiation as compared to higher doses. These dose-dependent differences are observed in both, protein coding and non-coding genes.
From the microarray analysis, 16 lncRNAs (those commonly regulated at the doses of 2 and 0.1 Gy) plus an additional set of 12 lncRNAs (from those exclusively regulated after 2 Gy of X-ray) were selected for lncRNA expression validation using qRT-PCR (Figure 6). To this end, two sets of primers were designed for each candidate, one set was designed to test the same RNA sequence as the one used in the microarray and the second set was intron-spanning which, considering that there are multiple gene annotations for each candidate, would allow a better approach to the proper gene sequence. Using this approach, we could finally confirm radiation-induced changes in the expression of 13 out of the 16 and 7 out of the 12 genes tested (Figure 6). In order to know what causes this incomplete validation, we examined RNA-seq results of breast luminal epithelial cells from the Tea Tlsty laboratory. In agreement with our results, none of the candidates that could not be validated by qRT-PCR showed reads in the RNA-seq analysis. Thus, imperfect annotation of lncRNAs that makes the probe design difficult might explain the observed partial inconsistency between microarray and qRT-PCR data. Anyway, we could finally validate 71.4% of genes identified with the microarrays, which is in the range of other studies carried out in many laboratories. Although understanding the role of lncRNAs in cellular response to radiation is still unexplored, our microarray and individual qRT-PCR assays indicate that the cellular response to DNA damage in terms of non-coding genome regulation depends on the radiation dose.

The radiation responsiveness of the Y-RNA long non-coding RNA.
The laboratory work involved cultivating a transformed breast cell line (04Bc) and testing the expression levels of the following genes relative to the endogenous TATA-binding protein (TBP) gene in cell lines (MDA-MB-361; MCF10A; 04Bc): Y-RNA (lincRNA also located in chrX: 21911718-21920026) and neighbouring protein coding genes CNKSR2 (Connector enhancer of kinase suppressor of ras2) and MBTPS2 (membrane bound transcription factor peptidase 2) located in proximity to chrX: 21911718-21920026. Focus was placed on linc Y-RNA rather than TCONS (8309bp) as an assay could be conveniently designed for this short ncRNA. LincY-RNA: Interestingly, linc Y-RNA revealed a contrasting radiation response expression profile at the 24hr time point in MDA-MB-361 compared to the normal breast cell line MCF10A (Figure 7). Increased expression of Y-RNA was evident at 4hr post low and medium dose irradiation in the transformed cell line 04Bc.

Functional study of a radiation-induced long noncoding RNA – SMAD5-AS1.
According to the microarray validation data, one of the non-coding genes with a higher fold change after DNA damage in a variety of cell types is SMAD5-AS1, which is an antisense lncRNA to SMAD5 protein-coding gene. Since we observed that SMAD5 was also positively regulated after damage induction, we investigated whether the lncRNA SMAD5-AS1 was acting as an enhancer for SMAD5. Differential extraction of RNA revealed that, while SMAD5-AS1 is mainly located at the cytosol, the SMAD5 RNA is mostly found in the nucleus. Thus, the lncRNA does not seem to regulate the expression of the protein coding gene.
In order to investigate the function of the lncRNA SMAD5-AS1, we proceeded to inhibit its expression. To this end, we designed short hairpin (sh)RNAs to knockdown the gene, and then cloned them into pLKO plasmids that were used to produce lentiviral vectors. MCF10A infected with the lentiviruses generated with some of the designed shRNAs showed a reduced expression of the SMAD5-AS1 while not significantly affecting the SMAD5 expression. The insensitivity of SMAD5 expression to the SMAD5-AS1 inhibition, together with the different cellular location of their corresponding RNAs indicates that the lncRNA does not participate in the regulation of the SMAD5 gene. To further decipher the biological function of SMAD5-AS1 transcript, we analysed proliferation, cell cycle progression, and cell death. The MTT assay revealed no differences in the proliferation rate of normal and knockdown cells after DNA damage induction. Similarly, Annexin V assay did not reveal differences in the number of apoptotic cells between the different samples. However, cell cycle analysis by flow cytometry revealed significant differences between normal and knockdown cells (Figure 8). Treatment of uninhibited cells with either doxorubicin or radiation leads to a significant increase in the fraction of cells in G2/M as a consequence of a strong checkpoint that impedes the division of cells with DNA damage. In contrast, cells with SMAD5-AS1 inhibited do not accumulate in G2/M, thus pointing to the involvement of the lncRNA in this checkpoint, which is of utmost importance in the cellular response to radiation-induced DNA damage.

The microRNA response to irradiation using a mouse model of radiation carcinogenesis.
The accumulation of DNA damage caused by IR in conjunction with disrupted cellular regulation processes can lead to carcinogenesis. However, radiation may also cause dysregulation of epigenetic processes that can lead to altered gene function and malignant transformation, and epigenetic alterations are important causes of miRNA dysregulation in cancer. miRNAs have been characterized as master regulators of the genome. As such, miRNAs are responsible for regulating almost every cellular pathway, including the DDR after ionizing radiation exposure. However, most miRNAs expressed in adults are tissue-specific and the pattern of miRNA target gene expression is complicated, especially in the central nervous system (CNS).
The cerebellar granule cells, the most abundant neurons within the entire mammalian CNS originate from granular cell precursors (GCPs), whose proliferation starts in the embryonic stage reaching maximum levels postnatally. This process is under strict genetic control and the Sonic Hedgehog (Shh) signalling pathway plays a major role. GCPs neoplastic transformation causes development of medulloblastoma, the most common CNS cancer in children. Previous studies by ENEA have revealed that the sensitivity of mice to radiation-induced medulloblastoma is dependent upon the normal functioning of the Ptch1 gene. This gene activates a growth and differentiation programme that is a strong candidate for regulation through the non-coding genome. Ptch1+/- mice are considered the best characterized murine model of medulloblastoma. In addition to tumour predisposition, these mice are characterized by high radiation susceptibility. For this reason they represent a useful model for studies of changes in the transcription profile of non-coding RNAs, particularly miRNAs, which occur during development of radiation-induced medulloblastoma.
The principal objective here was to compare microRNA and long non-coding RNA expression in mouse brains irradiated with low (0.01 to 0.1Gy ) and moderate (0.1 to 2Gy) X-ray doses.
Experiments were carried out using an ex vivo model in order to isolate CGPs populations from Ptch1+/- and wild type mouse cerebella in early postnatal age. To this aim, experimental conditions were standardized in terms of: i) number of animals to be used for each experimental point for isolating a sufficient number of CGPs; ii) cell culture conditions to ensure reproducible results. Isolated CGPs populations were then tested for expression of typical CGP markers and absence of neural differentiation markers. Total RNA was extracted from cultured non-irradiated or 1 Gy-irradiated GCPs of both genotypes, and next-generation sequencing was carried out. The list of miRNAs perturbed by radiation in WT- and Ptch1+/- GCPs were compared, and five common miRNAs were statistically significant as shown through the Venn plot in Figure 9. When comparison was carried out between the baseline levels of miRNAs expressed in WT- and Ptch1+/- GCPs in non-irradiated conditions and miRNAs levels 4h after irradiation of both genotypes, 8 common miRNAs were statistically significant.

A major achievement of this complex animal study is the identification, by bioinformatics analysis [i.e. Ingenuity Pathway Analysis (IPA)], of a subset of miRNAs, controlling different biological functions, whose expression was altered in GCPs by radiation alone or in combination with Shh-deregulation (i.e. miRNAs let-7a, mir-17, mir-34a, mir-144 and mir-486). These miRNAs are involved in DNA damage, inhibition of senescence and a concurrent increase of cell survival. The same miRNAs were validated in spontaneous and radio-induced MB from Ptch1+/- mice. Notably four out of five (with exclusion of mir-486) showed statistically significant up regulation or down regulation when their expression was compared in spontaneous and radiation-induced medulloblastoma. The most significant result of this task concerns the identification of a potential biomarker for radiation exposed organisms and radiation-induced tumours, i.e. miR-34. This conclusion is supported by a clear dose-dependent up regulation of this miRNA in irradiated GCPs (both genotypes) at short-term after irradiation (Figure 10) and, importantly, by a statistically significant decrease of miR34 expression in radiation-induced compared with spontaneous tumours.

Response of the long non-coding RNA TERRA to radiation.
Previous studies showed that reduced expression of Rb1 prevents radiation-induced cell cycle arrest and simultaneously creates shortening of the telomeres leading to an increase in genomic instability (Gonzalez-Vasconcellos et al. 2013). These findings were linked to the changes in expression of a telomere specific long non-coding RNA (lncRNA) called TERRA. Changes in such lncRNA, and therefore in the architecture of the telomeres, could explain the increase in susceptibility to radiation-induced cancer in Rb1 haploinsufficiency. Functional assays were established to investigate TERRA expression on human chromosomes 15, 10 and X/Y as well as on murine chromosomes 18, 10 and X/Y. Following the methodological validation, the expression of TERRA was studied on cells harbouring different Rb1 expression levels. We found that different levels of the Rb1 protein modulate the expression of the telomeric non coding RNA named TERRA. The TERRA minimal promoter unit of the human chromosome 15 was predicted in silico and cloned into a PGL3 luciferase expressing vector replacing its given promoter. Double and triple transfections into U2OS cells were performed with the TERRA vector (vTERRA) together with Rb1 modulating vectors. Data revealed that the Rb1 protein levels are crucial for the regulation of the TERRA promoter activity (Figure 11).
To study the telomere structural changes in Rb1 haploinsufficient osteoblasts, a new technique was designed (patent pending) to evaluate the differential epigenetic states of telomeres. In this new assay, we have combined a DNA nuclease digestion with an specific quantitative PCR (qPCR) of telomeric DNA, which we term the Telomere Chromatin Condensation Assay (TCCA). By quantifying the relative amounts of telomeric DNA that are progressively digested with the exonuclease Bal 31, the method can discriminate between different levels of telomeric chromatin condensation (Figure 12a). A telomeric characteristic histone modification study was performed together with the TCCA analysis (Figure 12b) to further assess the architecture changes in the Rb1 haploinsufficient cells.

Candidate radiation biomarkers of the non-coding transcriptome.
The HLEG recommendations for future research and the MELODI transitional research agenda both recommend the use of radiation biomarkers. These are seen as a tool to distinguish between tissues developing disease pathologies due to exposure to radiation and those developing the same pathologies due to other (idiopathic) causes such as ageing or other environmental noxae. Although several pathways and endpoints have been suggested to hold potential as radiation biomarkers (Pernot et al, 2012) none have yet been shown to distinguish exposed and non-exposed cases. It is our contention that non-coding RNA species, due to their abundance and persistence, may offer some potential as biomarkers. Indeed they are used in many situations to detect tumours and to predict cancer therapy outcomes.
A second problem with the use of biomarkers is the availability of biological material from subjects exposed to radiation. Many key epidemiological cohorts involve individuals who are no longer available to provide materials, where it is impractical to obtain the required ethical approval for sample collection, or where individual dosimetry is not available. The Serbian Tinea Capitis Cohort consists of individuals with long-term health outcomes, whose clinical records are available and where individuals can be identified and contacted for sampling. Moreover the highly standardized dosimetry and availability of non-exposed siblings makes this a potentially useful cohort.
A third problem with biomarker analysis is the lack of validated protocols that allow individual samples to be collected and handled in a reproducible manner. Within this problem complex the issues of storage, tracking and transport must also be considered.

Development of a standard operating procedure for sample collection.
After considerable testing of materials it was decided to use the following standardized protocol for collection: One PAX-gene Tube, one PLASMA Tube and one EDTA Blood Tube were transferred to Munich in April 2015 for RNA isolation and analysis of defined ncRNA biomarkers. The rest (2 PLASMA Tubes, 1 EDTA Blood Tube and 2 URINE Tubes) are stored in repository at -80°C in IMGGE (Belgrade). The flow chart for collection and storage is presented in Figure 13.

Feasibility study (collection and transportation of samples) for use in retrospective biological analysis.
Preparatory discussions and logistical planning meetings were held during the Dark.Risk 2nd Annual Meeting in Belgrade (October 2014). It was concluded that 50 samples planned for the feasibility study should include 30 cases and 20 controls. The field test of the collection included: (1) approaching and in-depth interviewing of 514 individuals treated as children with X-irradiation (2) cross checking these person in the Cancer Registry and (3) taking biological samples (venous blood and urine) Biological samples were taken, transported to IMGGE (Belgrade), processed and stored at -80°C until shipment to HMGU.
The results of the study were that it proved completely possible to obtain sufficient samples. The goal of 30 exposed and 20 general controls was achieved within a 6 month period that included tracking each subject through local and central records, visiting them and collecting samples.

Validation of SOP by analysis of non-coding RNA levels in biological samples.
The quality of the collected biological material was determined upon delivery to Germany. DNA and RNA isolation was performed using an automated programme (Maxwell16 Promega). The concentration of RNA isolated from plasma samples was very low (5-10ng/µl) but was adequate to allow PCR amplification of low abundance target RNAs. We analysed non-coding RNAs identified in the Dark.Risk project using qRT-PCR miRNA assays to quantify expression (Figure 14). Detected (Ct) values for RNU-U6 (endogenous control) were between 30-35 cycles and for miR-21 and miR-16 between 25-30 cycles.

Potential Impact:
Impact on optimizing protection, improving regulation and on clinical practice

Dark.Risk will make a contribution to the optimisation of the protection afforded to radiation workers, patients undergoing medical procedures involving radiation, and the general public. This will come in the long-term through improvements in regulatory processes. This impact will come from future molecular epidemiology studies, through new knowledge of the shape of the dose response curve and through recognition of differences in individual sensitivity to radiation.

Article 31 of the EURATOM treaty and Council Directive 96/29/EURATOM of 13 May 1996 mandates that the community adopt basic standards to protect both radiation workers and the population from adverse health effects arising from exposure to ionizing radiation. The current system of radiation protection for the protection of workers, and the public exposed to therapeutic/diagnostic procedures, as discussed by advisory bodies such as ICRP, BEIR and UNSCEAR, does not consider the influence of individual susceptibility or the possible non-linearity of the dose response curve when drafting standards. Dark.Risk addresses these issues by establishing the Serbian Cohort of Tinea Capitis Children for future molecular epidemiological studies on the risks of late effects of radiation.

The improved knowledge of the biological mechanisms gained from identifying the non-coding transcriptome as a key radiation response will allow the construction of more accurate biologically-based mathematical models, which will in turn better inform on the shape of the dose response relationship.

Dark.Risk research will provide new information on the long-term health effects of radiation exposure in children as the subjects were all pre-pubertal at the time of exposure. This will be vital in making future evidence based judgments on the potential risks and benefits of radiation exposures in the young.

Impact on European radiation research, mobilisation of National funding and added value
The maintenance of academic and technical competence in Europe has become a key issue recognised by MELODI. Dark.Risk has increased European competence by recruiting and training junior scientists in the radiation sciences. Also by establishing new co-operations between epidemiologists and radiobiologists we have opened new possibilities for long-term interdisciplinary cooperation. We have established research activities in new partner institutes (SERGAS, ROSA) and created a new resource (SRTCC). This will provide opportunities for the consortium partners to attract additional national funding for their continued research.

The European Research Area has historically played a major role in shaping radiation protection legislation, with major contributions from European scientists to international organisations such as NEA, ICRP and UNSCEAR. Through the activities of Dark.Risk we have established a new voice in national radiation protection decision-making in Spain and Serbia, as well as strengthening the expertise in Germany and Italy. The Dark.risk consortium has established a new European platform of radiation research (SRTCC). This would not have been possible under national funding. The continuing cooperation between the individual laboratories will ensure that there is a sustained transfer of knowledge

Project dissemination has been primarily via publications in the open scientific literature, presentations at national and international conferences. An additional successful activity has been the incorporation of the new knowledge in teaching activities (Bachelor, Masters, and Doctoral levels) in each of the partner member states. The project has presented its results to other EURATOM consortia and has included other consortia at each of the Dark.Risk annual meetings.

List of Websites: