University of Calgary PRISM: University of Calgary's Digital Repository
Graduate Studies The Vault: Electronic Theses and Dissertations
2018-12-17 An Analysis of Human Exposure to Alpha Particle Radiation
Stanley, Fintan
Stanley, F. (2018). An Analysis of Human Exposure to Alpha Particle Radiation (Unpublished doctoral thesis). University of Calgary, Calgary, AB. http://hdl.handle.net/1880/109368 doctoral thesis
University of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission. Downloaded from PRISM: https://prism.ucalgary.ca UNIVERSITY OF CALGARY
An Analysis of Human Exposure to Alpha Particle Radiation
by
Fintan Stanley
A THESIS SUBMITTED TO THE FACULTY OF GRADUATE STUDIES IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
GRADUATE PROGRAM IN BIOCHEMISTRY AND MOLECULAR BIOLOGY
CALGARY, ALBERTA
DECEMBER, 2018
© Fintan Stanley 2018 Abstract High linear energy transfer (LET) ionizing radiation (IR) is the predominant source of IR humans are exposed to. Radon gas, which emits a high energy alpha-particle, represents the greatest single lifetime source, but also remains comparatively understudied versus low LET IR sources such as x-rays. The inhalation radon (222Rn) gas from indoor air exposes lung tissue to alpha particle radiation, damaging DNA and increasing the lifetime risk of lung cancer. Buildings can concentrate radioactive radon (222Rn) gas to harmful levels. To enable cancer prevention, I examined how Canadian Prairie radon exposure is modified by environmental design and human behavior and evaluated different radon test modalities. I also developed a high-throughput, benchtop alpha-particle irradiation system to facilitate future research into the biological consequences of high LET radiation exposure. Initially, I examined 90+ day radon test results from 2,382 residential homes from an area encompassing 82.5% of the Southern Alberta population. Remediated homes were retested to determine efficacy of radon reduction techniques in this region. Subsequently, 11,726 Alberta and Saskatchewan homes were radon tested, coupled to geographic, design and behavior metrics. Canadian Prairie homes contained 140 Bq/m3 average radon (min <15 Bq/m3; max 7,199 Bq/m3) and 17.8% were ≥ 200 Bq/m3. Geostatistical analysis indicates significant variation between regions. More recently constructed homes contain higher radon versus older. Finally, I also designed and validated a benchtop, 96 well plate-based 241Am irradiation system to expose cultured eukaryotic cells to alpha particles in a controlled environment. My validation of this novel setup includes quantification of nuclear alpha particle-induced DNA damage signalling (γH2AX) using a purpose-designed 3D analysis method, physical readouts of alpha particle-induced DNA damage by alkaline comet assay, and an investigation of cellular viability after alpha particle exposure. This method brings significant advances over existing techniques in its ease of setup and use, affordability, accessibility and flexibility and should enable future alpha particle radiation biology. Collectively, my work demonstrates that radon is a genuine public health concern in the Canadian Prairies, legitimatizes efforts to understand the consequences of radon exposure to the public, and suggest that radon testing and mitigation is likely to be an impactful cancer prevention strategy.
ii
Preface
Certain introductory material in this thesis has also been peer-reviewed and published as: Berger ND, Stanley FKT, Moore S, Goodarzi AA. ATM-dependent pathways of chromatin remodelling and oxidative DNA damage responses. Philosophical Transactions B. 2017 Oct 5;372(1731). Invited Review. Stanley FKT, Moore S, Goodarzi AA. CHD chromatin remodelling enzymes and the DNA damage response. Mutat Res. 2013 Oct;750(1-2):31-44. Invited Review. Moore S, Stanley FKT, Goodarzi AA. The repair of environmentally relevant DNA double strand breaks caused by high linear energy transfer irradiation - no simple task. DNA Repair. 2014. 17:64-73. Invited Review.
The work detailed in Chapter two of this thesis has been peer-reviewed and published as: F.K.T. Stanley, S. Zarezadeh, C.D. Dumais, K. Dumais, R. MacQueen, F. Clement, & A.A. Goodarzi, “Comprehensive survey of household radon gas levels and risk factors in southern Alberta.”. CMAJ open, 2017, vol. 5, issue 1, E255-E264
The remainder of the thesis is original, unpublished, independent work by the author, Fintan K.T. Stanley
The experiments reported in Chapters 2-3 were covered by Ethics Certificate number REB17-2239, issued by the University of Calgary Conjoint Health Ethics Board for the project “The Albertan ‘Evict Radon’ Radon Awareness and Testing Campaign” on December 15, 2016.
iii Acknowledgements My thanks to Dr. Aaron Goodarzi for his mentorship and guidance throughout my studies. Thank you for fostering such a broad and rewarding experience during this time. Thanks also to Dr. Susan Lees-Miller & Dr. Jen Cobb for years of generous advice and support.
I warmly thank the members of the lab past and present for their support. My thanks to the many students I had the honor to attempt to share some knowledge with, teaching always reconnected me to the wonder of science. For keeping me close to that thank you, and I hope you all take a little of that wonder with you in your many successful journeys.
I greatly appreciate all the hours of help and support from those at the Arnie Charbonneau Cancer Institute, particular thanks to Carmen, Carin, Anne, and most especially Shilpa.
Thank you to the Department of Biochemistry and Molecular Biology whose dedication to the students’ experience is a testament to the quality of its leadership, thank you Sarah and Jonathan, and special thanks to Leslie.
I thank all my collaborators, no matter what came of the work, I most expressly thank you for engaging in co-operative science and promoting a more altruistic approach in research. Most distinct thanks here to Dan Berger who has been inspiring advocate for this philosophy. For the many ideas that bore fruit at the bench and the multiples more that did not, thank you for helping me test and refine them.
And with the deepest deference I thank my friends, I feel too lucky to have half of you in my life. Your belief in me got me through this, and I shall endeavour second to live up to your ideal of me, and first to remind you each, as often as I can, what wonderful generous people you are. Thank you UREKA. Thank you Heather. Thank you Dan.
Thank you all so very much & also thank you Canada!
iv Dedication
For my family,
With special thanks to my parents, Audrey and Bob, for always supporting my wandering,
And to my brothers and sisters, for always welcoming me home again.
v
Table of Contents
Abstract...... ii
Preface ...... iii Acknowledgements ...... iv Dedication...... v
Table of Contents ...... vi
List of Tables ...... xi
List of Figures and Illustrations ...... xii List of Symbols, Abbreviations and Nomenclature ...... xiv Epigraph ...... xvii
Chapter One: Introduction ...... 18 1.1 A Short History of the Discovery of Ionizing Radiation ...... 18 1.2 Radiobiology meets DNA in the 20th century ...... 21 1.3 Radiation and Cancer – DNA damage is the missing link ...... 25 1.3.2 In Brief: DNA Damage Signalling and Repair Pathways ...... 27 1.3.3. In Detail: IR-induced DNA Double Strand Break Signalling ...... 30 1.3.4. In Detail: IR-induced DNA Double Strand Break Repair Pathway Choice .. 32 1.4.5. Factors Influencing DSB Repair Kinetics ...... 33 1.4 Radiation Quality, Linear Energy Transfer and Relative Biological Effects ...... 35 1.5 ICRP models in the context of DNA damage ...... 37 1.6 Defining Clustered DNA damage and implications for DSB repair ...... 39 1.7 Radon Inhalation – The Most Common Form of Human IR Exposure ...... 43 1.8 Other Sources of High LET IR Exposure ...... 46 1.10 Hypothesis and Specific Aims ...... 48
Chapter Two: Comprehensive Survey of Household Radon gas Levels and Risk Factors in Southern Alberta...... 50
vi 2.1 Preface ...... 50 2.2 Abstract ...... 51 2.3 Introduction ...... 52 2.4 Methods ...... 53 2.4.1 Setting and Design ...... 53 2.4.2 Data Collection ...... 53 2.4.3 Statistical Analysis ...... 54 2.5 Results ...... 55 2.5.1 Household Radon level in the Calgary metropolitan are ...... 55 2.5.2 Radon by home feature metrics ...... 57 2.5.3 Radon by home age and mitigation status ...... 58 2.6 Interpretation ...... 59 2.6.1 Main Findings ...... 59 2.6.2 Explanation and Comparison ...... 60 2.6.3 Strengths and limitations ...... 61 2.6.4 Implications ...... 62 2.7 Conclusion ...... 62
Chapter Three: Environmental design and behavioral variables influencing radon gas exposure and testing accuracy in the Canadian Prairies ...... 67 3.1 Introduction ...... 67 3.2 Methods ...... 68 3.2.1 Main Study Setting and Design ...... 68 3.2.2 Radon Test Comparison Subsets ...... 69 3.2.3 Data Collection ...... 69 3.2.4 Geospatial Analysis ...... 70 3.2.5 Statistical Analysis ...... 71 3.3 71 3.4 Results ...... 71 3.4.1 Extent, Structure, and Overall statistics of the Study ...... 71 3.4.2 Geographical Analysis of the Alberta and Saskatchewan Radon Testing ...... 72
vii 3.4.3 Household Radon Concentrations as a Function of Year of Building Construction ...... 77 3.4.4 Household Radon Concentration as a Function of Square Footage and Ceiling Height ...... 79 3.4.5 The impact of occupant behaviour, other home metrics and the overall model ...... 82 3.4.5.1 Radon as a function of structural attributes of the basement level and foundation ...... 82 3.4.5.2 Occupant behaviour influencing home air dynamics ...... 84 3.4.5.3 Overall and Subset regression Models ...... 88 3.4.6 Examining short term (5 Day) radon testing ...... 89 3.5 Discussion ...... 103 3.5.1 Summary / Interpretation ...... 103 3.5.1.1 Build Types ...... 103 3.5.1.2 Differences in Radon Concentration Levels by Year of Construction 105 3.5.1.3 Geospatial Differences in Residential Radon Levels ...... 106 3.5.1.4 Metrics of Environmental Design Dimensions ...... 106 3.5.1.5 Occupant Behaviours and its impact on Radon Levels ...... 107 3.5.1.6 Overall model ...... 108
Chapter Four: High-throughput, benchtop alpha particle irradiation system for the investigation of high linear energy transfer radiation in biological systems ...... 109 4.1 Introduction ...... 109 4.2 Results ...... 111 4.2.1 Design and dose quantifications of the alpha particle irradiation system ..... 111 4.2.2 γH2AX induction, high-resolution image acquisition and data analysis pipeline ...... 114 4.2.3 α particle irradiation induced DSB repair kinetics ...... 118 4.2.4 Quantification of high LET α particle irradiation induced DNA damage by alkaline comet assay, and effects on cell viability ...... 121 4.2.4.1 Alkaline comet assay ...... 121 4.2.4.2 Alamar blue cell viability assay ...... 123
viii 4.2.5 Screening potential and adaptability to other genetic model systems ...... 125 4.3 Discussion ...... 127
Chapter Six: Interpretation ...... 130 6.1 Main Findings ...... 130 6.2 Explanation and Comparison ...... 130 6.3 Strengths and limitations ...... 131 6.4 Consequences ...... 134
Chapter Seven: Materials and Methods ...... 137 7.1 Mammalian Cell Culture Work ...... 137 7.1.1 Cell lines and Tissue Culture ...... 137 7.1.2 Transfections ...... 137 7.2 Reagents ...... 137 7.2.1 Antibodies ...... 137 7.2.2 Inhibitors & Compounds ...... 138 7.2.3 Solutions & Buffers ...... 138 7.3 Alpha/Gamma IR and 241Am Source Dosimetry ...... 139 7.4 Design and 3D Printing ...... 140 7.5 Immunofluorescence ...... 140 7.6 Widefield Epifluorescence Microscope Imaging ...... 141 7.7 Image Analysis ...... 141 7.8 Confocal Microscope Imaging ...... 141 7.9 Statistical Analysis ...... 142 7.10 Alkaline comet assay ...... 142 7.11 Alamar Blue cell viability assay ...... 142 7.12 Cell Lysate Preparations and Western Blotting ...... 143 7.12.1 Immunoprecipitation...... 143 7.13 UV Laser Microirradiation and Live Cell Imaging ...... 144
References ...... 145
Appendix A - Defining the relocalization of CHD5 to DNA damage sites ...... 166
ix 10.1 Preface ...... 166 10.2 Introduction ...... 167 10.2.1 DNA double strand break repair and chromatin ...... 167 10.2.2 CHD5 ...... 168 10.3 Results ...... 170 10.3.1 Endogenous CHD5 expression is difficult to demonstrate and antibody specificity could not be demonstrated...... 170 10.3.2 CHD5 knockdown did not impact DSB repair kinetics in human cell lines172 10.3.3 CHD5 does recruit to laser microirradiation tracks, but no dependency for this process were found ...... 173 10.4 Conclusions ...... 180
Appendix B – Cell cycle staging based on DAPI...... 182
Appendix C - Alpha and gamma irradiated A549 cells ...... 183
Appendix D – Chronic low dose alpha irradiation of yeast cells ...... 184
Appendix E - Real time radon data ...... 185
Appendix F – Geospatially divided radon results ...... 186
Appendix G – BILL 209...... 195
x List of Tables Table 2-1- Radon levels by home metrics ...... 64
Table 2-2 - Linear Model of Radon Levels Based on Home Metrics ...... 65
Table 3-1 - Radon Levels by Location of Testing ...... 94
Table 3-2 - Radon Levels by Geographic divisions based on ED and major population centers ...... 95
Table 3-3 - Radon Levels by Year of Construction ...... 95
Table 3-4 - Radon levels by Home Dimensions ...... 96
Table 3-5 - Radon levels by Home Dimensions ...... 97
Table 3-6 - Radon levels by Basement Attributes ...... 98
Table 3-7 - Radon levels by Thermostat Settings ...... 99
Table 3-8 - Radon Concentration by Frequency of Window Opening ...... 100
Table 3-9 – Post hoc test (pairwise reporting) for radon level by home metric ANOVAs 101
xi List of Figures and Illustrations Figure 1-1 Condensed timeline of major events in radiation and DNA research 1895- 1955 ...... 19
Figure 1-2 – DSB response: ATM signaling axis ...... 29
Figure 1-3 - Measures of Radiation and Linear Energy Transfer ...... 36
Figure 2-1- Indoor air radon concentrations by postal code district in the greater Calgary metropolitan area...... 55
Figure 2-2 - Average indoor air radon concentrations by subdivision of the greater Calgary metropolitan area ...... 56
Figure 2-3 - Average indoor air radon concentrations by home features ...... 57
Figure 2-4 - Maximum observed household air radon concentrations are significantly higher in homes built in the past 25 years...... 58
Figure 2-5 - Pre- and post-mitigation household air radon concentrations for all homes initially measuring ≥200 Bq/m3...... 59
Figure 3-1 – Extent of Residential Radon testing in Alberta and Saskatchewan...... 73
Figure 3-2 – Residential radon testing results divided by testing location with the residence...... 74
Figure 3-3 – Map showing radon testing statistics across Alberta and Saskatchewan...... 75
Figure 3-4 – Full spread of all residential radon tests, also divided by geographic region . 76
Figure 3-5 – Residential radon concentrations divided by year of home construction...... 78
Figure 3-6 – Residential radon testing results divided by major dimensions of the building...... 80
xii Figure 3-7 – Residential radon testing results divided by Ceiling Heights of respective floors ...... 81
Figure 3-8– Residential radon testing results by attributes of basement (i) ...... 82
Figure 3-9 – Residential radon testing results by attributes of basement (ii) ...... 83
Figure 3-10 – Residential radon testing results across different thermostat setting behaviours ...... 86
Figure 3-11 – Residential radon testing results across different frequency of window opening ...... 87
Figure 3-12 – Subset selection in overall regression analysis ...... 88
Figure 3-13 – Long- vs Short- term testing, separated by testing location in home ...... 90
Figure 3-14 – Long- vs Short- term testing, separated by short term test date ...... 91
Figure 4-1 – Design and Dosimetry of 241Am Irradiation setup ...... 113
Figure 4-2 – Example Images of alpha particle irradiation ...... 114
Figure 4-3 – Automated image analysis, workflow and examples ...... 116
Figure 4-4 – Image analysis Outputs ...... 117
Figure 4-5 – 48Br cells DSBR in an asynchronously population after alpha IR compared to gamma IR...... 120
Figure 4-6 – Alternative Setup for Alamar and Comet assays ...... 121
Figure 4-7 – Adaptations of the assay: Comet, Alamar Blue and Inhibitors ...... 124
Figure A-1 Diagram of the amino acid sequence of CHD5 highlighting main protein domains ...... 168
xiii List of Symbols, Abbreviations and Nomenclature
Symbol Definition A549 Lung Adenocarcinoma Cell Line Alt-NHEJ Alternative Non-homologous End Joining ANOVA Analysis of Variance AP Apurinic/apyrimidinic AT Ataxia Telangiectasia ATM Ataxia Telangiectasia Mutated ATMi ATM inhibitor: KU55933 ATP Adenosine Triphosphate ATR Ataxia telangiectasia and Rad3 related BEIR Biological Effects of lionizing Radiation BEIR Biologic Effects of Ionizing Radiation BER Base Excision Repair BIC Bayesian Information Criterion BRCA1/2 Breast Cancer Type ½ BrdU 5-Bromo-2-‘-deoxyuridine BSA Bovine Serum Albumin Cdk Cyclin-dependent kinase CHD Chromodomain Helicase DNA-binding CHD3 Chromodomain Helicase DNA Binding Protein 3 CHK1/2 Checkpoint kinase 1 or 2 CMV Cytomegalovirus C-NRPP Canadian National Radon Protection Program CNS Central Nervous System CtIP CtBP-Interacting Protein DAPI 4’6-diamidino-2-phenylindole DDR DNA Damage Response DMSO Dimethyl Sulfoxide DNA Deoxyribonucleic Acid DNA Deoxyribonucleic Acid DNA-PK DNA-dependent Protein Kinase DNA-PKcs DNA-dependent Protein Kinase Catalytic Subunit DSB Double Strand Break DSBR Double Strand Break Response dsDNA double stranded DNA ECL Enhanced Chemiluminescence EGFR Epidermal Growth Factor FDA Food and Drug Authority FISH Fluorescence in situ Hybridization
xiv FSA Forward Sortation Area GCR Galactic Cosmic Rays GFP Green Fluorescent Protein ɣ-H2AX Phosphorylated histone H2A (phospho-S139) GLM General Linear Model HAT Histone Acetyl-Transferase HC Health Canada HDAC Histone DeAcetyl-Transferase HeLa Human cervical adenocarcinoma HR Homologous Recombination HZE high (H) atomic number (Z) and energy (E) IARC International Agency for Research on Cancer ICL Interstrand Crosslink ICRP International Commission on Radiological Protection IF Immunofluorescence IP Immunoprecipitation IR Ionizing Radiation IRIF Ionizing Radiation Induced Foci IUPAC International Union of Pure and Applied Chemistry KAP-1 KRAB-associated Protein 1 LD50 Lethal Dose 50 LET Linear Energy Transfer MEF Mouse Embryonic Fibroblast MEF Mouse Embryonic Fibroblasts MMR Mismatch Repair MRN Mre11, Rad50, Nbs1 NB neuroblastoma NCI National Cancer Centre NER Nucleotide Excision Repair NHEJ Non-Homologous End Joining NuRD Nucleosome Remodeling and Deacetylase PAR Poly(ADP-ribose) PARG Poly(ADP-ribose) Glycohydrolase PARP Poly(ADP-ribose) Polymerase PARPi PARP inhibitor (olaparib) PBS Phosphate Buffered Saline PFA Paraformaldehyde PFGE Pulse Field Gel Electrophoresis PIKK phosphatidyl inositol 3-kinase like protein kinase PKcsi DNA-PKcs inhibitor: NU7441 PNKP Polynucleotide Kinase/Phosphatase PTM Post-Translational Modification
xv RBE Relative Biological Effectiveness ROS Reactive Oxygen Species RPA Replication Protein A RT Room Temperature SAM S-adenyl Methionine SAXS Small Angle X-ray Scattering SB Sample Buffer SEM Standard Error of the Mean SEP Solar Energy Particles shRNA Small hairpin RNA siRNA Small interfering RNA SSB Single Strand Break SSBR Single Strand Break Repair ssDNA single stranded DNA TANGO Tools for Analysis of Nuclear Genome Organization U2OS Human Osteocarcoma cell line UN United Nations UV Ultraviolet UVA Ultraviolet A UVLM UVA Laser Microirradiation WHO World Health Organization XRCC I/III/IV X-ray Repair Cross-Complementing Protein
xvi Epigraph
Built from destruction You pull it all apart Make yourself an ending Call yourself a start
xvii Chapter One: Introduction
1.1 A Short History of the Discovery of Ionizing Radiation
Early studies into radiation focused on the visible light spectrum, being a form of radiation that humans can be presumed to have always known about. Radiation is the transfer of energy in either the form of electromagnetic rays or high-speed particles Studies in diffraction and the use of prisms led to the discovery of the various forms of invisible radiation, first infrared (William Herschel) and then ultraviolet (UV) (Johann Wilhelm Ritter) in the early 1800s. The end of the 1800s brought significant progress in the study of radiation, when the rapid advancement of electrical engineering allowed for the artificial production of radiation. It was in this period that radio waves (Heinrich Hertz) and X-rays (Wilhelm Röntgen) were first characterised(fig1-1). Rontgen’s X-rays could penetrate certain materials and cause fluorescence of barium platinocyanide, with the ‘X’ signifying the common nomenclature for an unknown1. Rӧntgen famously produced the first X-ray image of his wife’s hand and wedding band, but also described the burns that occurred (on the hands) during the X-ray photography. Although he mistakenly attributed this to free radicals produced in the air by the X-rays, this likely represents the earliest documentation of biological effects of ionizing radiation (IR). In the year after Rӧntgen’s publication, Henri Becquerel discovered that certain materials spontaneously emitted similar rays 2,3. Marie Curie then showed that this was due to a specific set of chemical elements and ascribed the word ‘radioactivity’ to this property4(fig1-1). Towards the end of the 19th century, Ernest Rutherford described two other types of IR differing from X-rays, distinguishing them based on penetrative ability. He characterised these as alpha radiation (which had very low penetrative capacity) and beta radiation (whose penetrative ability was intermediate between alpha and X-ray), designations we still use today5. A few years later, once a sufficiently strong magnet had been developed to bend the beam of an alpha particle, Rutherford showed that alpha and beta particles were ionised particles, and thus described the earliest characterisation of particle radiation6.
18 The earliest known descriptions of the element radon also come from this era. Becquerel characterised a radioactive gas emitted by thorium-containing minerals, which was initially termed “thorium emanation” (this was actually 220Rn). Friedrich Ernst Dorn, in his explorations of the thorium emanations characterised by Rutherford, investigated the radioactive emanations of several compounds. This included radium, where he found analogous emanations, which he imaginatively called the “radium emanation” (this was
Figure 1-1 Condensed timeline of major events in radiation and DNA research 1895-1955
19 actually 222Rn). The Curies also appear to have been aware of the emanations from thorium and radium in 1899, and in a 1901 paper by Rutherford and Harriet Brooks on emanations, they credited the Curies for the discovery. A. Debierne observed similar emanations from actinium, which he termed the “actinium emanation” (this was actually 219Rn)7. Rutherford and F. Soddy showed the thorium emanation was an inert noble and radioactive gas8,9. In 1923, the International Committee for Chemical Elements and International Union of Pure and Applied Chemistry (IUPAC), which had itself been established in 1919, clarified the naming nomenclature. The names radon (Rn), thoron (Tn), and actinon (An) were ascribed to the various radon isotopes. Later, when all radioisotopes were numbered instead of named, the element took its primary name from the most stable isotope, radon, and the less stable isotopes were assigned accordingly: 222 220 219 Rn (radon, t1/2 = 3.8 days), Rn (thoron, t1/2 = 55.6 seconds), and Rn (actinon, t1/2 = 3.9 seconds). In 1909, Paul Villard described another form of ray emanating from radium, and Rutherford subsequently called this a gamma-ray, completing the ontology of radioactive decay. Other forms of radioactive decay have been discovered since those early days of radiation science, including positrons and neutrinos however, the α, β and γ designation remains commonly and widely used today. A major breakthrough in the field was achieved in 1914 when Rutherford and Andrade showed that an electrified vacuum produced x-rays that were related to gamma rays and differed in wavelength10. During this time, work by Victor Hess also showed that IR decreased up to an altitude of 1 km from the Earth’s surface, but then began to increase (up to 5.3 km)11. This new source of IR was proposed to be penetrating the atmosphere from outer space and represents what we now know as “cosmic rays”, although that specific term was coined decades later by Robert Milken, as he confirmed and expanded on Hess’s work12. It was Arthur Compton who, in the 1930s, correctly characterised cosmic rays as being composed of ions13. Another point where meteorological research greatly informed radiation research was through the work of C.T.R. Wilson. From 1896 to 1912, Wilson developed the cloud chamber and it became the first method capable of visualising ionizing radiation14,15(fig1- 1). The “rain-like” condensation observed after exposure of the clouds to the newly discovered x-rays fitted with observations made by JJ Thomson and JA McClelland after
20 Rӧntgen’s discovery, that is to say air could be made conductive by X-rays16,17. Thomson and Rutherford confirmed that the conductivity was due to ionization of the gas, and it was confirmed that the ion chamber could be used to detect and photograph the ionisation tracks17. The cloud chamber remained an important method for the study of IR for decades, and they are the earliest predecessor of the modern silicon particle detectors that, which when arranged around a collision in a particle accelerator, give accurate detection of subsequent particle paths. Parallel to these endeavours, others began to use the x-rays as a tool for the visualisation of crystals. In 1912, A. Sommerfeld reported the successful diffraction of X- rays by MV Laue, P Knipping and W Friedrich18. This research was greatly expanded by WH Bragg who, by 1921, was using x-rays to study the structure of organic crystals 19. Bragg is also noteworthy for his 1903 characterisation of the energy loss of ionizing radiation as its travel through matter, a plot of this is known as a Bragg curve, and the characteristic maxima on these plots is known as a Bragg peak (this phenomenon is today exploited in proton beam therapy to minimise off-target dose deposition). His son, WL Bragg, continued work in x-ray crystallography and derived Bragg’s Law which remains vital to x-ray crystallography20(fig1-1).
1.2 Radiobiology meets DNA in the 20th century
DNA was first isolated from surgical bandages and was called nuclein, as it resided in the nucleus of the cell. This discovery, by Freich Meiesher in 1869, was followed by Albrecht Kossel’s isolation of the non-protein components and the characterisation of the 5 nucleobases: A (adenine), G (guanosine), C (cytosine), T (thymine) and U (uracil)21(fig1-1). In 1938, following parallel breakthroughs in radiation research, W Asbury and F Bell used X-ray diffraction techniques to show that DNA had a regular, repeating structure22. It was then, in 1953 following the crystallisation and x-ray diffraction imaging of DNA by Rosalind Franklin and Raymond Gosling, that Francis Crick and James Watson proposed the now-accepted double helix model of DNA23(fig1- 1). In the years between the first crystallisation (1869) and the 1953 Nature paper describing DNA, the field of genetics had begun to incorporate findings and techniques
21 from radiation research, leading to the 1927 presentation by Herman Muller of his paper “The Problem of Genetic Modification” in which he showed a quantitative connection between IR exposure and lethal mutations in Drosophila melanogaster 24,25. Muller’s observation of the mutagenic effect of X-rays lead to him being the chief proponent of the ‘linear no-threshold’ model of IR exposure. This model is central to modern radiobiology and is the basis of almost all widely operating policy on human radiation protection26,27. The early 20th century was a time of great public interest in radiation and its possible implications for biology and medicine. The era saw the broad sale and consumption of radiation-based medicines, many of which contained the element radium, which had been discovered by Marie & Pierre Curie in 1898. Walter Lazarus-Barlow established that radium accumulated in the bone and, by 1914, a review of more than 500 medical case reports by Ernest Zuebline documented bone necrosis and ulceration as side effects of radium consumption28,29. Despite this, such products were still sold until the late 1930s. In the late 1920s, Eben Byers developed cancer and ultimately died after consuming purportedly large amounts of radium, so much so that he was buried in a lead- lined coffin. At the same time, thousands of factory workers developed radiation poisoning from the paint used on watch dials that also contained radium. Both cases received considerable public attention, the latter leading to numerous litigations. The Byers case is reported to have spurred campaigns to expand the jurisdiction of the FDA, and the Food, Drug, and Cosmetic Act (FD&C Act) was signed into law in the United States on June 24th, 1938, significantly increasing regulatory authority over drugs by mandating a pre-market review of the safety30. The new law and the dial worker lawsuits lead to the vast reduction of radium exposures. Study of the exposed populations continued in the following decades, and a thorough review of these populations was published by the Argonne national laboratory that closely followed radium-exposed individuals31.
In studying the radium clock dial workers, Robley Evans, a seminal figure in nuclear medicine, developed a method to measure the accumulation of radium in the body32,33. Studies by Evans, Aub, Flinn, & Martland during this period represent some of the first epidemiological studies of radiation exposure in humans and established some of
22 the first guidelines for human IR exposure. In 1941, Evans proposed a body burden of 0.01 µCi as “tolerance level” for human radioprotection (NBS 1941; NCRP 1941). These findings and standards were later reviewed by Evans (1981), where he stated “the proper subject for the study of man is man” referring to his earlier studies attempting to model exposure in rats (1944), where he found as much as 150x times the radium in the bones of rats was required to see the chronic effects observed in humans. In 1945, towards the end of World War II, the atomic bombing of the cities of Hiroshima and Nagasaki by USA military forces resulted in the IR exposure of a large population on an unprecedented scale. Efforts to treat survivors lead to the documentation of “atomic bomb sickness’ now known as acute radiation syndrome. In the years that followed, the Atomic Bomb Casualty Commission (later the Radiation Effects Research Foundation) would be established to follow these populations to continue to document and treat the effects in the population. The Life Span Study of atomic bomb survivors continues to study the population and is perhaps one of the most informative studies epidemiological studies of human radiation exposure due to its size and duration. The solid cancer dose response derived from is evidence in support of the linear relationship between human radiation exposure and cancer risk34–36. The watch dial workers case had shown that occupational exposure to radiation was a serious public health concern. An important figure pioneering considerable work in the field of occupational radiation exposure and resulting diseases was Wilhelm Hueper. During his tenure as the first director of Environmental Cancer Section of the National Cancer Institute (NCI), he published major reviews on occupational exposures and cancer, and he was an early advocate against asbestos37–39. European mining in the Erz Mountains was likely the earliest example of an occupation associated with an increased risk of lung cancer, and this was first reported in a 1879 report of some 650 miners, which gave clinical and autopsy descriptions of intra-thoracic neoplasms40. The 150 reported deaths from "miner's disease" were probably from lung cancer. As early as the 1940s, Hueper warned that the increased incidence of lung cancer amongst miners was likely due to the radiation exposure caused by radon inhalation in the mines. Although at the time, others didn’t believe that the amount of radiation resulting from radon would be sufficient to explain the cancer levels, the missing piece of information then was the considerable
23 radiation emanating from radon progeny (other radioactive elements along its decay series)38,41,42. There were numerous cohort studies carried out on miners across Europe and North America in the following decades, and these showed that occupational exposure to radon decay products was associated with an increased risk of lung cancer (see NBK304363 for a more comprehensive review43). In 1998, an important meta-analysis was carried out to attempt to increase statistical power of the models. BEIR IV combined analysis from American Canadian and Swedish studies. Subsequent analysis covering a total of 11 cohorts developed risk models that considered the concentration and duration of the miners’ exposure (BEIR VI 1999)44, and also notably BEIR IV held to the linear “no threshold” model (as does the most recent BEIR VII (2006))36. Two major studies were subsequently released in 2006, a German cohort study of 59,000 uranium miners and a Canadian meta-analysis of 17,660 uranium miners (Beaverlodge, Port Radium, Port Hope), and the patterns of lung cancer risk from each of these studies was in line with those of BEIR VI. Although it is true that today a uranium miner may realistically receive a greater dose of IR from radon in their home versus their work environment, it wasn’t until 1967 that Canada and the US set radon exposure limits in uranium mines45. These miner studies have been central to the development of models for radioprotection and our understanding of the radon and lung cancer risk association. This greater understanding was also essential for subsequent investigation into residential radon exposure which are discussed later (section 1.8) There are multiple other historical examples of large population scale exposure to IR. These include the 1957 Sellafield (“Windscale”) nuclear waste facility fire, the 1979 Three Mile Island Nuclear reactor coolant release, the 1986 Chernobyl nuclear power plant meltdown, and the 2011 tsunami-hit Fukushima Daiichi nuclear power plant meltdown. These incidents and their health effects have been extensively reviewed over the years and yet, in almost all cases, estimates of additional cancer diagnosis and mortality resulting from the radiological release have varied. Briefly, in most cases the IR exposure was studied in two cohorts: the workers involved in the cleanup operations (who may be expected to have rather high and acute doses), and the general population (who may subsequently expose to any released radioisotopes); estimates in both of these groups are contested. For example, amongst Chernobyl cleanup workers 134 people were
24 hospitalised for acute radiation poisoning in the aftermath. Of these, 28 died in the following months, with 14 more IR-attributable deaths over the following years. However, for the cleanup workers who received lower doses, some reports show no statistically significant increase in malignancies, while others do46,47. WHO estimates for the number of additional thyroid cancer deaths caused by the release of radioactive iodine isotopes stand at only nine. In a total estimate of IR-induced cancer deaths post the Chernobyl incident, WHO looked at the 200,000 emergency workers, 116,000 evacuees and 270,000 residents of the most contaminated areas and estimated an additional 4,000 cancer deaths48. Excluding the effects of the Chernobyl disaster, 25% of that population would be otherwise expected to die from cancer, thus 4,000 additional deaths would equate to 3% rise, which may prove difficult to measure (cardis et al. 2006). In any study of IR exposure related to cancer incidence, the long latency period of cancer and individual differences in cancer susceptibility (based on genetics and confounding environmental variables) means that these studies take decades to reach firm conclusions. Hence, the continued monitoring of IR affected populations remains an on-going endeavour in all the above cases. These events remind us to remain vigilant to the risks of radiation exposure, however, it is frequently noted in many of these studies that the mental health burden (so called “radiophobia”) can far out way the cancer risk burden in exposed populations. As such, it remains important to not exaggerate the effects of IR exposure47,48.
1.3 Radiation and Cancer – DNA damage is the missing link
Genomic instability is now considered an enabling characteristic of cancer 49, based on continued work in the fields of radiation biology, the biochemistry of DNA and cancer genetics. It is now accepted knowledge that genetic mutations are the driving events in cancer formation, cancer evolution, anti-cancer therapeutic resistance, and that IR is readily capable of inducing these driving changes in DNA structure. Despite increasing understanding and ever-improving survival rates, cancer is still a major disease burden. In Canada, it is the single greatest cause of death, and Canadians have a 1 in 2 chance of being diagnosed with cancer in their lifetime, with a 1 in 4 chance of dying from the disease 50. For healthcare reasons, considerable investments have been made into DNA
25 damage and repair research as they relate to sources of cancer in our environment, the treatment of cancer and the long-term consequences of anti-cancer therapy.
1.3.1 Types of DNA Damage Cells regularly experience DNA damage, not just from exogenous sources such as IR but also from the chemical reactions that occur endogenously as a part of normal cellular metabolism. The burden of DNA damage that cells encounter takes the form of a huge variety of chemical modifications of the DNA double-helix. These result in the genetic information becoming unreadable or misread, severely limiting cellular functionality and threatening a cell’s ability to replicate. It is in the evolutionary interest of cells within a multicellular organism to repair any damage before any corrupted information is used to produce erroneous proteins or is replicated and propagated into daughter cells. In fact, human cells are remarkably efficient at repairing DNA damage, despite some estimates placing the number of DNA lesions per day as high as 20,000 51. Within a normal cell, the overall mutation rate is incredibly low and, even accounting for baseline error rates in DNA replication, there is on average less than 1 mutation that modifies a protein per generation (0.35 amino acid coding mutations per generation)52. However, as each form of damage represents a structurally-distinct obstacle to the cell, they require a wide range of enzymes to repair them. Endogenous DNA damage is typically spontaneous, such as base depurination or cytosine deamination that may result in AP (apurininc/apyrimidinic) sites or base transitions, respectively 53,54. Damage can also occur via reactions with metabolic co- factors such as SAM (S-Adenosyl methionine) or with the by-products of incomplete respiration, i.e. ROS (Reactive Oxygen Species). These forms of damage can result in several chemically-altered bases with different base pairing properties to the undamaged forms. Exogenous damage can occur through exposure to UV, genotoxic chemicals or IR. UV damage results in pyrimidine dimers and many of the most prevalent genotoxins (such as those found in cigarettes) produce aromatic (bulky) adducts. Other chemicals can cause ICLs (Interstrand Cross Links). IR results in breaks in one or both strands of the DNA deoxyribophosphate backbone, triggering DNA single strand breaks (SSBs) or double strand breaks (DSBs) 55. Even more frequent than direct ionization of DNA is IR-
26 induced radiolysis of water molecules to generate ROS such as hydroxyl radicals, superoxides (and more) that subsequently elicit oxidative DNA damage.
1.3.2 In Brief: DNA Damage Signalling and Repair Pathways A network of DNA repair pathways has evolved to deal with the many distinct forms of DNA damage encountered by the cell. Incorrectly matched bases are repaired by MMR (Mismatch Repair), and the smaller chemical alterations to DNA bases are repaired via BER (Base Excision Repair). Larger chemical alterations are repaired by the removal and re-synthesis of a ~30 base oligonucleotide in NER (Nucleotide Excision Repair). ICLs are repaired by a complex set of pathways broadly classified as interstrand cross link repair 56. Both directly-induced SSBs and SSBs that arise as repair intermediates of other pathways are repaired by single strand break repair 57. Notably, DNA damage of almost any kind can be worsened during DNA replication, explaining why replicative and/or young tissues are particularly sensitive to IR. This is due to the often catastrophic collision of replication forks progressing along a strand of DNA with DNA lesions that lead to destabilization of the fork, first stalling fork progression and potentially leading to full collapse 58,59. Stalled or collapsed replication forks are a significant risk to the cell as they can potentially result in failure to separate sister chromatids, mitotic catastrophe and cell death. DSBs are a particularly dangerous form of DNA damage and eukaryotes have evolved multiple redundant repair mechanisms. These are often characterised as distinct repair pathways, although the seemingly precise divisions between pathways represent our ontological construction more than reality of the cellular milieu. DSBs result in a complex response that can ultimately trigger to one of two repair mechanisms: HR (Homologous Recombination) or NHEJ (Non-Homologous End Joining). HR is generally reserved for S and G2 phase, where a sister chromatid is readily available, whilst NHEJ is the main form of repair at all other times and still functional in S and G2. DNA repair pathways encompass a vast network, some of which are constitutively active and respond to commonly occurring adducts, while others are only activated in response to specific damage types during specific phases of the cell cycle. One of the major questions in DNA Damage Response (DDR) research is how a cell controls this vast network and “makes decisions” as to which pathway is most appropriate and when.
27 Indeed, the DDR must be precisely regulated, as the effectors of this pathway include a vast array of enzymes capable of manipulating the DNA helix, changing the base sequence and altering the chromatin structure 56. Any mis-regulation of the DDR network could cause severe genomic instability and, indeed, many of the components within DDR pathways were identified in mutant cells displaying moderate to severe genomic instability. Many of these observations came from screens of human cells from families with genetic diseases of genome instability and cancer predisposition, which were ultimately identified to be caused by mutations in the DDR gene network. The DDR is a highly coordinated and powerful cellular network which responds to a wide range of genomic infringements and develops a response within the context of the nature of the damage, the surrounding chromatin state, the cell cycle status and undoubtedly any other number of homeostatic markers. Protein recruitment is central to the function of the DDR, with sensor proteins eliciting a process that can recruit sufficient DDR factors so that microscopically visible foci develop 60. These discrete nuclear structures are a commonly used as experimental markers to demarcate DNA damage induction and active DDR signalling. It is believed that the exact kinetics of recruitment, the order and abundance of recruited proteins has a significant control over the output of DDR signalling. The temporal control of protein recruitment may also have a significant consequence for the ultimate cellular outcomes of the damage 61. PTM (Post Translational Modifications) has been shown to be a crucial player in foci development. Phosphorylation, ubiquitination, SUMOylation and PARylation (Poly ADP-ribosylation) all occur in response to DNA damage. Most of the PTMs are recognised by specific protein domains which mediate the protein recruitment 56. To highlight the complexity of DDR signalling, I will use the response to DSBs as a case example. The DSB response is centrally mediated by members of the phosphatidyl inositol 3-kinase like protein kinase (PIKK) family, which includes (amongst others): ATM (Ataxia Telangiectasia Mutated), ATR (Ataxia Telangiectasia and Rad3-related protein) and DNA-PKcs (DNA Protein Kinase, catalytic subunit) 59. ATM is involved in the response to DSBs and has a vast substrate network. DNA-PKcs has a much narrower group of validated substrates, limited more to the regulation of end processing and DSB end joining in NHEJ. ATR is most heavily involved in stalled replication fork signalling
28 and DSB signalling after end-resection. There is a large overlap in the substrate specificity between ATM and ATR, but while there is a degree of redundancy within the network, the two proteins also mediate independent events 62. The ATM and ATR kinases are involved in the early events of the DDR and, after the detection of lesions by sensor proteins, ATM and ATR phosphorylate mediator proteins that act primarily as platforms for the recruitment of other ATM/ATR substrates. These recruited factors often serve to
Figure 1-2 – DSB response: ATM signaling axis ATM can be activated in the context of a DSB by the MRN complex or by direct oxidation. In response to a DSB, MRN binds DSB ends, and promotes the autophosphorylation and activation of ATMS1981p. When activated, ATM generates ɣ-H2AX Chk2T68p and p53S15p. This promotes recruitment of downstream signaling and effector proteins, including 53BP1, which interacts with ATM and effectors such as RIF1, PTIP and SCAI. In response to oxidative stress, inactive ATM homodimers are directly oxidized and activated, bypassing MRN-dependent autophosphorylation. adapted from Berger et al. ATM-dependent pathways of chromatin remodelling and oxidative DNA damage responses. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 372, (2017).
29 re-enforce and amplify DDR signalling, while other proteins can remodel the surrounding chromatin to facilitate or further enhance signalling and enable access for repair enzymes. I have previously reviewed the interface between the DSB response and chromatin extensively and refer readers to that article for a more comprehensive discussion of this topic63.
1.3.3. In Detail: IR-induced DNA Double Strand Break Signalling As discussed above, DSBs arise naturally because of endogenously produced ROS- mediated DNA damage, the stalling and collapse of DNA replication forks, and/or exposure to IR or radiomimetic chemicals. In all cases, the DNA phosphodiester backbone is severed on opposing strands at two proximal locations where the intervening base-pairing is insufficient to hold chromosomes together. With even one DSB being potentially lethal for a replicating cell, there are robust mechanisms and extensive redundancy within cells to both detect and repair these lesions. Indeed, DSBs are extremely dangerous lesions as they are particularly prone to cause chromosomal rearrangements and genomic instability 64. There are four known sensors of DSBs: (i) the Ku70/80 heterodimer, which promotes NHEJ; (ii) PARP, which competes with Ku70/80 and promotes alternative (alt)-NHEJ; (iii) the Mre11–Rad50–NBS1 (MRN) complex, which promotes end resection, revealing 3’ ssDNA overhang which is required for homology searching in SSA and HR; and (iv) Replication Protein A (RPA), which binds ssDNA and is involved in the recruitment of SSA and HR effector proteins, RAD52 and RAD51 respectively65. DSB response signalling ATM and ATR are the key protein kinases activated following DSB detection(fig1-2). ATM responds to double-stranded DNA ends bound by the MRN complex and ATR is activated by DSBs with long single-strand DNA overhangs coated in RPA, such as occurs following replication fork collapse or during DNA end-resection initiated during HR-mediated repair. Both ATM and ATR belong to the PIKK family, which shows a strong preference for phosphorylating serine or threonine followed by glutamine (S/T-Q) and includes DNA-PKcs, a key component of the NHEJ pathway of DSB repair66. One substrate targeted by all three kinases is S139 of the histone variant H2AX60. H2AXS139p (usually referred to as γH2AX) appears within
30 minutes of DSB formation, spreading across ∼1 megabase (Mb) of chromatin to either side of the lesion67. Following irradiation, the primary kinase that targets H2AXS139 is ATM, with DNA-PKcs capable of modifying H2AXS139 in the very immediate vicinity of the DSB, and ATR only at those breaks resulting from collapsed replication forks or that have been resected. While γH2AX is not thought, in and of itself, to have a direct impact on nucleosome spacing, it serves as the nucleating factor for a hierarchical series of events that collectively promote an enormous alteration to the chromatin surrounding the DSB60. γH2AX has become a convenient and powerful tool to monitor IR-induced DSBs68. An early target of the PIKK protein kinases during DSB detection and signalling, γH2AX nucleates the subsequent accumulation of numerous other chromatin-binding and modifying factors (including MDC1, RNF8, RNF168, 53BP1 and RIF1), forming so- called IR-induced foci (IRIF)69. IRIF can be visualised using immunofluorescence microscopy, with their relative number over time serving as a surrogate measure for DSB induction and repair68. After H2AX phosphorylation, MDC1 specifically binds γH2AX and facilitates the subsequent accumulation of the RING-finger ubiquitin ligases, RNF8 and RNF168, whose activity towards histone H2A then enables BRCA1 and 53BP1 retention on chromatin surrounding the DSB 70–72 (fig1-2). Importantly, the binding of MDC1 to γH2AX initiates the amplification of DSB signalling, whereby more and more MRN is concentrated at DSBs (via interactions with either MDC1 or 53BP1) to activate more and more ATM (via ATMS1981 autophosphorylation), thus expanding the damage signal into surrounding areas to form IRIF that spread around the DSB site73–77. Even more proteins regulating IRIF formation have been described in the past decade; these include RNF16978,79 and HERC280,81, which regulate the Ubiquitin (Ub)-ligase activity of RNF8/RNF168, a cadre of enzymes regulating DSB responsive histone SUMOylation, including RNF482,83, PIAS1 and PIAS484 and several proteins that accumulate at DSBs downstream of 53BP1/BRCA1 retention, including RIF185–87, EXPAND188 and others. Most of these IRIF proteins can be used to demarcate DSBs using fluorescent microscopy; γH2AX foci have been of great use to elucidate of the mechanisms and kinetics of DSB repair89.
31
1.3.4. In Detail: IR-induced DNA Double Strand Break Repair Pathway Choice To help maintain the genetic code against such damage, life has evolved a complex network of pathways which provides the cells with no less the four ways to repair DSBs: NHEJ, Alt-NHEJ, SSA or HR. These processes occur within the single network and it is currently understood that cell uses one pathway when the others have failed or are unfeasible. While the term “pathway choice” is often used in referring to how lesions ultimately get repaired, it is important to maintain that the cell is incapable of choice or decision-making (as we might typically use those phrases), and that all of this is mediated by the stochastic interactions of biomolecules. HR is the least error prone form of DSB repair as it relies on the re-synthesis of any lost or intervening information using inter-chromosomal homology to template the repair. Typically, this occurs using a sister chromatid to repair the damage but, for this same reason, HR is largely restricted to G2 and S phases of the cell cycle where the homologous partner is available. NHEJ lies at the other end of the DSB repair spectrum, as it constitutes the direct re-joining of two DSB ends with minimal end processing and lacks any need for homology. Alt-NHEJ, sometimes known as micro homology mediated end joining (MMEJ), relies on shorter less accurate homology between two DSBs to match them for repairing, and requires a greater degree of end processing than NHEJ, but nothing as significant as that seen in HR. SSA is a process where, if a DSB occurs between units of a repeating genomic element, the homology of the repeats can be used to re-anneal the two opposing strands. This process requires the repeats to be in parallel and results in the loss of intervening information. DSB processing involves the removal of an oligonucleotide from on strand at the end of the DSB, resulting in a 3’ ssDNA overhang. This overhang can then be used in the search for homology59. ATM is perhaps the most central protein in the DSB response, and it has many substrates known to play a role in DSB repair pathway choice 62. It is believed that, initially, the Ku heterodimer binds the DSB ends, and that DNA-PKcs plays a role in stabilising the break ends. Phosphorylation events are then used to control whether the dsDNA ends are protected from processing enzymes, which typically favour NHEJ- mediated repair. Specific phosphorylation events on the DNA-PKcs six residue “ABCDE
32 cluster” counter the end protection events, and destabilize he DNA-PKcs interaction with dsDNA ends90. ATM can induce this phosphorylation to promote HR, typically in cases where NHEJ has failed. In some cases, the DSB end must be processed before it can be re-ligated in NHEJ. This process involves several enzymes, including Artemis/SNM1C and PNKP, the latter of which re-establishes the correct 3’ and 5’ phosphorylation state of DSB termini so that relegation is energetically and enzymatically feasible 91. PARP can also detect DSBs and competes with Ku for DSB end binding. PARP promotes alt-NHEJ, which acts as a “backup” to classical NHEJ 92. PARP is also thought to mediate the recruitment of factors like MRN and ATM to sites of damage, as demonstrated by the delayed phosphorylation of ATM substrates in PARP1-deficient cells 93. ATM and MRN then instigate the classical γH2AX DDR cascade94. After PARP activation, the MRN complex can accumulate at sites of damage and mediate end processing, preparing the DNA for HR and promoting ATM activation and subsequent activation of resection factors such as CtIP, BRCA1 and MRN95. Resected DSB ends contain the newly exposed ssDNA that is bound by RPA, which is a critical player in the activation of ATR signalling via ATRIP 96,97. In the case of repetitive sequence elements, these RPA-coated ssDNA can be annealed by RAD52-mediated SSA. Alternatively, RAD51 filaments may replace RPA on the overhang via BRCA2, and this is often considered the beginning of canonical HR 98. The importance of ATM in HR is supported by evidence from ATM-deficient cells which display recombination defects. If HR or NHEJ are unavailable or unable to resolve DSBs, processes such as ‘alternative NHEJ’ and/or single-strand annealing may also repair damage, albeit with greater risk of error (for a comprehensive review of all these pathways, see 59). Current models indicate that the orderly progression of DNA repair is mediated by the successive inhibition of the previous (failed) repair pathway by the subsequent one. ATM is involved in most levels of this “decision making” and its primary role is often seen to be the mediating of successive recruitment of multiple rounds of DDR factors to the sites of damage.
1.4.5. Factors Influencing DSB Repair Kinetics How the chromatin surrounding the DSB impacts repair pathways is an emerging area of research, with significant portions of the mechanism of NHEJ and HR, and the choice
33 between the two, being reliant on substantial alterations to nucleosome spacing. To appreciate this, it is important to review the following basics of each mechanism: NHEJ is initiated by binding of the Ku70/80 heterodimer to DSBs, which then recruits DNA-PKcs whose autophosphorylation regulates access to the DSB termini and thus DNA end- processing events. Once DSB termini are processed (into a ligatable form), the DNA Ligase IV–XRCC4–XLF complex physically rejoins the ends. NHEJ works quickly, resolving lesions within minutes to hours, and requires no homology between ends. Should NHEJ fail (or be unsuited, such as in the case of collapsed DNA replication forks), an ATM-dependent process initiates DNA end resection to generate 3’ DNA overhangs coated in RPA, giving way to ATR signalling, the BRCA2-dependent formation of Rad51 filaments and subsequent strand-invasion of the undamaged template and repair by HR. Given these added steps, HR can take many hours to complete and has an absolute requirement for a homologous template. Current evidence suggests that the DSB end processing phase of NHEJ, as well as the DNA end resection and Rad51 filament formation stages of HR, are hugely dependent upon upstream chromatin remodelling events that are often specific to the chromatin context within which the DSB occurs99,100. Indeed, differential DSB response kinetics, magnitudes, pathway choices and genetic requirements for repair have been found to resolve lesions occurring within euchromatin versus heterochromatin, with even greater intricacy of response found when one examines each highly specific context101. Broadly speaking, however, euchromatic DSBs are detected and repaired according to the canonical processes and kinetics described above, while DSBs occurring within highly compacted or more complex chromatin structures (e.g. pericentric heterochromatic chromocenters, telomeric shelterin complexes, nucleoli, etc.) generally require longer and/or additional factors for resolution. This phenomenon is readily observed using IRIF enumeration over time, where foci (representing DSBs) triggered by X-rays or γ-rays disappear with at least two kinetically distinct components: approximately 80–90% are repaired with fast kinetics and 10–20% more slowly68,69,102. Observed in G0, G1 and G2 phase, the slow component of DSB repair requires can occur by either NHEJ (G0, G1) or HR (G2)68,76,103–105. It was the study of ATM, the γH2AX→53BP1 signalling hierarchy and Artemis-deficient cells that demonstrated that
34 this slow, ATM-dependent fraction of repair encompassed DSBs within heterochromatic regions enriched for repressive histone mark H3K9me3 and the transcriptional co-repressor KAP-1 (KRAB-ZFP-Associated Protein 1, also called TIF1 or TRIM28 or KRIP1)104. Without ATM, relaxation of the restrictive heterochromatic superstructure failed and DSBs within these areas persisted104,106. Studies from yeast and other systems also find that heterochromatin is a barrier to DSB signalling, naturally limiting the spread of γH2AX and, in the most condensed regions, requiring dynamic movement of DSBs to the heterochromatin periphery for IRIF formation to occur67,107–109. It was demonstrated that the somatic mutation rate of cancer cells is substantially higher within H3K9me3, H3K9me2 or H4K20me3-rich (heterochromatic) regions, versus any other epigenetic histone modifier110. It is possible that the higher mutation rates within heterochromatin observed in this study reflect the difficulties in DSB signalling, repair or even an increased exposure to mutagens at the periphery of the nuclear structure, where, in many cells, heterochromatin is selectively located. 1.4 Radiation Quality, Linear Energy Transfer and Relative Biological Effects
Briefly some relevant units to radiation and how it is measured; Becquerels (Bq)represents one radioactive decay event per second, this often coupled to an area measurement to produce Bq/m3 readings. Gray is unit of absorbed dose, it represents the radioactive energy absorbed by an object and equates as 1Gy = 1 joule. The Sievert is a measure of equivalent dose it a measure of the gray dose that corrects for the type of radiation and the tissue absorbing it (discussed further below), ICRP calculation equate 1 Sv to 5.5% increase in cancer risk (fig 1-3). As discussed previously, the types of DNA damage that cells experience is chemically very diverse and, even within the narrower realm of IR exposure, a wide variety of chemical damage can occur. This variety is, to some degree, dependent on the physical characteristics of the IR, which can itself vary considerably across the electromagnetic spectrum and the many forms of particle radiation. Linear Energy Transfer (LET) is an important way of comparing different radiation types. Measured in keV/μm (Newtons), and sometimes referred to as stopping power, LET is an expression of the energy loss of the radiation over unit distance (fig 1-3). Of course, this is energy is not ‘lost’, rather it is transferred of transformed, and to biologically contextualise that, we must consider how
35 Figure 1-3 - Measures of Radiation and Linear Energy Transfer this energy might be transferred to DNA (or the water and protein solution in which it sits), leading to ionization and chemical damage. So, higher-LET radiation will lose its energy over a shorter distance, but this is not a measure of dose, rather it characterizes the unique dose deposition features of that particular type of IR. IR LET can be calculated in two main different ways, restricted LET and unrestricted LET. As the name implies, LET assumes in the linearity of the radiation; however, in reality IR can displace electrons along its path, and these may scatter out from the linear path of the radiation (this is commonly referred to as delta radiation) and, depending on the energies involved, may travel far and result in further ionisation events along their own paths. It is important to realise that these are complex events, and that LET is a not a
36 perfect modelling of events. For example, LET is often presented as an average, such as a single number that may be quoted for the LET of an alpha particle emitted by decaying radon isotopes. Alpha particles have an energy of 5.49 MeV and a range in air of 414 mm; in this case, the LET quoted for such a model is in fact an average of the LET which varies along the length of the track (Bragg peak) (fig 1-3). Also, while gamma-rays are termed low-LET, and are typically presented as the quintessential example of this, strictly speaking the term has no relevance to the photons moving at light speed comprising the gamma ray, as they interact with matter in a starkly different way compared to heavy ions travelling at ‘only’ half the speed of light. LET values that are quoted for electromagnetic radiation are usually unrestricted LETs and, on whole, largely describe the effects of the delta radiation as opposed to the photon itself, this approximation serves, as it is the delta radiation that causes the ionisation events we are most interested in from a biological effect perspective. One definition that was used to describe the biological effects of different qualities of IR is the relative biological effectiveness (RBE), an estimated constant for each “quality’ of radiation that, when factored with dose, calculates the corresponding biological effect.
Experimentally, these are most often the result of LD50 calculations from cells grown in laboratory culture conditions, and the relevance of these numbers to human whole model radiation or tissue specific effects must be estimated. It might be clear now to the reader that are a great deal of estimations and assumptions that go into models of the biological effects of IR. This is one reason to continue the work, so that the estimates can be tested against real data. To synthesize the current operating model, we can look to the work of international organisations that work closely on these calculations. The International Commission on Radiation Protection (ICRP) is a United Nations (UN) sanctioned body that has worked for decades to synthesize and translate the research in this area, with their models being widely adopted as the basis for national radiation protection programs26. Except where otherwise stated, I will keep closely to the ICRP models.
1.5 ICRP models in the context of DNA damage In ICRP 26, a model of RBE was built based on the unrestricted LET of different types of IR in water. However, in ICRP 60 this model was largely replaced by a series of assigned
37 radiation weighting factors (Wr) based on the type and energy of the radiation and can be used to calculate an “equivalent dose” in Sieverts (Sv) based on the dose absorbed (Gy). ICRP also suggests calculation for an “effective dose”, which includes tissue specific sensitivities and produces a tissue weighted sum of equivalent doses. Effective dose is the most widely accepted quantity used in radiation protection. The unit of effective does is the Sievert, which is an SI derived unit, receiving a 1 Sv equivalent dose of radiation consequences a 5.5% probability of developing cancer. That probability is calculated within the assumptions of the linear no threshold model. The 2007 ICRP report 26 continues to use the assigned radiation weighting factors as a measure of RBE, although they are not calculated based on LET. The system provides conservative estimates while removing much of the complexity of these calculations; in this system, the Wr of x-rays and gamma rays are set to a relative value of 1, whilst alpha particles and all heavy ions are assigned a Wr of 20. ICRP models are designed to provide useable and conservative models for radioprotection; however, they are greatly abstracted from the minute details of how the different forms of radiation differ in the chemical and spatial modification of DNA. On this scale, we continue to find LET as a useful concept for exploring radiobiology. To return to this scale of discussion; a classic DSB is often represented by clean dsDNA ends with an easily ligatable 5’-phosphate and a 3’-hydroxyl terminus. While IR may produce many such lesions, they likely represent the minority of events, and more ‘complex’ or ‘dirty’ breaks may be more common. Complex breaks are a catch-all term but may include altered DSB termini (5’-hydroxyls, 3’-phosphates, phosphoglycolates, protein adducts, etc.) or surrounding base alterations (AP sites, intra-strand & inter-strand crosslinks, etc.) and the frequency of complex lesions has been shown to increase with LET111. This considers the chemical diversity of the lesions, but their spatial diversity is also relevant – the phenomenon of clustering or the relative proximity of one DSB (or other lesions) to another. Higher LET IR results in more lesion clustering due to the confined nature of the energy deposition, alpha particles also happen to deposit ∼90% of their energy within a 10 nm radius along the linear track, producing DSBs spaced only about 10–20 bp apart112,113. Ionization events along the phosphodiester DNA backbone can generate closely spaced DSBs114. Water ionization can generate ROS that may also
38 react with bases or the phosphodiester backbone to generate yet more DSBs in the local area114. While lower LET IR will also produce the same spectrum of lesions, these will be more widely spaced, and may only have a small number of nearby crosslinks or base damages in close enough proximity to interfere with DSB repair113,114.
1.6 Defining Clustered DNA damage and implications for DSB repair Clustered DNA damage is defined as two or more lesions formed within one or two helical turns of DNA caused by the passage of a single radiation track115–118. Simulation systems have shown that high-LET IR can create as many as 25 lesions per 1–2 helical turns119. This finding is supported by studies modelling DNA damage following traversal by varying IR particles, which suggest that 70% of DSBs are clustered with other DSBs for high-LET IR, dropping to only 30% clustered lesion induction in cells hit with low- LET IR113,114,120. If other types of DNA damage (anything other than a DSB) are accounted for, then >90% of high-LET IR-induced DSBs are considered clustered, versus 60% for low-LET IR120. AP sites were found to arise most frequently within 8–10 bp of a DSB end121, whilst ICLs are clustered with a DSB with a comparably lower frequency, perhaps because they form only where mismatched bases or another non-hybridized DNA are present and ionized111,113,122. DSBs induced by high-LET α-particles are known to have a slower rate of repair compared to DSBs caused by low-LET γ-radiation123–125. Pulse-field gel electrophoresis (PFGE) experiments suggest that 90% of DSBs are repaired 3 h after 10 Gy γ -rays, contrasting with only 50% of DSBs being resolved 3 h after an equivalent dose of α particles. Indeed, 30% of α-particle induced-DSBs remained at 24 h post irradiation, with 2–5% still present up to 10 days later126. Greater cellular radiosensitivity following α- particle exposure is observed by clonogenic survival assay in comparison to γ -rays127. Although, dose-for-dose, both high and low-LET IR may produce similar DSB numbers, the closely packed damage caused by high-LET IR is more likely to result in small double stranded DNA (dsDNA) fragments compared to low-LET IR128,129. Increasing amounts of small dsDNA were observed by PFGE following irradiation with charged particles of increasing LET and were not observed after low-LET IR130.
39 The generation of short dsDNA fragments may be problematic to DSB repair for several reasons: (i) competitive interactions with DSB signalling and repair factors131, (ii) possible diffusion away from sites of rejoining (deletion of genetic information), (iii) loss of associated nucleosomes and (iv) a propensity for triggering mutagenizing events such as gene amplification132. In agreement with loss of dsDNA stretches from clustered DSBs, numerous studies have found that genetic mutation frequency increases with increasing LET of the IR in question133,134. Mutagenesis following high LET IR is further exacerbated as clustered DSBs are more likely to be mis-joined to one another due to the extreme proximity of multiple DSB ends; indeed, 50% of α-particle-induced DSBs repaired within 24 h are rejoined incorrectly126. FISH analysis has shown that simple exchanges between two chromosomes arise from a single point of damage, and therefore are linear with dose135. Complex chromosome aberrations (>3 breaks within ∼2 chromosomes) are readily induced following exposure to high-LET IR, even at very low doses. FISH studies determined that 83% of chromosome exchanges following low-dose α-particle exposure are ‘complex’, compared to only 30% for high dose X-irradiation112. Chromatin organization, specifically heterochromatinization (as indicated by the epigenetic marks H3K9me3 and H4K20me3), has also been demonstrated to influence regional mutation rates in multiple types of cancer136,137. Heterochromatinization is also a well-established modifier of IR-induced DSB repair, suppressing DNA end-joining unless nucleosomes are relaxed through cell signalling processes104,138. In non-dividing cells, IRIF analysis after heavy (iron) ion exposure indicates that DSBs are repaired with similar biphasic kinetics to DSBs induced by lower LET IR, albeit with a larger fraction of slow-repairing, remaining heterochromatic foci persisting 24–48 h post IR139. This data suggests that a fraction HZE induced DNA damage may be irreparable, possibly a result of other clustered damage which complicates repair and increases mutagenicity139,140. Notably, γH2AX foci were found to be an average of 1.5 times larger following high-LET IR compared to low-LET IR, possibly due to the clustering of multiple damage sites along the track into single IRIF141. γH2AX foci clustering occurred in only a subset of α- particle tracks and most γH2AX that clustered did so within 15–60 min of induction142. The issue of clustering has complicated the analysis of DSB repair following high-LET IR exposure, with considerable technical difficulty in determining accurately the number
40 of IRIF present within a dense track or highly clustered focus142–145 and individual γH2AX foci were difficult to post HZE exposure145. Based on broad evidence for the need for chromatin pathways in effective DNA repair I undertook an investigation into a chromatin remodelling protein and its potential role in DSB repair after irradiation (appendix A). IRIF may also be observed in cells not traversed directly by a high-LET IR track, explained by the potential ‘crossfire’ of -particles (delta radiation) into neighbouring cells. A recent report has suggested that, for HZE radiation has been reported alongside a less intense γH2AX nuclear distributed signal146. This pan-nuclear signal was not associated with apoptosis, occurred in multiple cell lines and increased in intensity with increasing dose. The conclusion from this interesting study was that heavy-ion associated pan-nuclear staining is caused by dispersed ATM and DNA-PKcs activity triggered away from the track of particle traversal146. It is possible that this distant activity could be associated with the production of small dsDNA fragments or through the ionization events produced by delta radiation.
1.7 A brief synopsis of NHEJ and its role in high-LET IR-induced DSB repair There is evidence for slower DSB rejoining by NHEJ following high-LET IR exposure compared to low-LET IR. The percentage of DSBs repaired by the slow component of NHEJ increases five-fold in cells exposed to nitrogen ions compared to gamma rays125; increasing LET correlated with a more unjoined DSBs present147. One group compared DSB repair kinetics and DSB end resection in G1 or G2 phase following exposure to carbon ions, X-rays or the topoisomerase poison, etoposide, finding that carbon ion- induced DSBs were resolved with the slowest kinetics in either G1 or G2 phase102. A delay in rejoining was also observed in another study, for cells irradiated with high-LET carbon and iron ions, compared to low-LET X-rays148. In most cases, delays in NHEJ- mediated DSB rejoining were attributed to increased lesion clustering directly reducing repair pathway efficiency Both Ku70 and Ku80 are thought essential for NHEJ and, Ku-deficient cells irradiated with low-LET IR show radiosensitivity and profound defects in DSB repair149– 151. By contrast, high-LET IR fails to kill more Ku80−/− rodent cells relative to the same
41 dose of low-LET IR, and no defect in DSB resolution is seen in Ku80−/− cells compared to the control152. It has been reported that >40 bp of ‘free’ dsDNA is required to load a Ku heterodimer on the end of a single dsDNA fragment, and that the minimum length of DNA for Ku-binding is 14 bp153. Competitive inhibition of NHEJ by an excess of free dsDNA fragments has also been investigated in vitro using increasing concentrations of duplex competitor oligonucleotides of varying size. NHEJ was found to be most sensitive to 42 bp DSB fragment inhibition, indicating this could be the minimum size that may permit the binding of the complete NHEJ complex131. High-LET exposure produces short DSB fragments of ∼40 bp152. Hence, one hypothesis is that such fragments interfere with the productive binding of Ku to chromosomal DSB ends, whilst not negatively impacting PARP-dependent alternative NHEJ or Mre11-dependent HR, neither of which is influenced strongly by small dsDNA fragments for activation154. Indeed, high-LET IR was found to kill significantly more wild-type and even greater numbers of HR-deficient mice (Rad54−/−) relative to low-LET IR. Strikingly, Ku70−/− mice were no more sensitive to high-LET IR than low-LET IR, all dying within 10 days of exposure154. While elevated radiosensitivity is observed in DNA-PKcs deficient rodent cell lines exposed to low-LET IR, no difference was observed compared to the control when the DNA-PKcs−/− cells were exposed to high-LET IR154155. Similar results were also obtained for DNA-PKcs positive and negative human cell lines, both showing profound but similar sensitivity towards high-LET IR156. Like Ku-dependent NHEJ, it was suggested that clustered lesions generated by high-LET IR may be a fundamentally inhibitory environment for DNA-PKcs-mediated end-joining155. This is supported by the phenomena of high-LET IR DNA fragmentation. With high-LET IR, the occurenece of clustered DSBs associated with DNA fragments could be inhibitory to DNA-PKcs activity and Ku-binding, and may render the DNA-PK holoenzyme an ineffective player in the ‘normal’ high LET IR induced DSB repair process. In agreement with this notion, DNA-PKcs autophosphorylation is significantly delayed following exposure to heavy (carbon or iron) ions compared to X-rays, suggesting an inhibition in protein kinase activation or ability to productively trans-autophosphorylate 148. One possibility is that the abundance of small Ku-bound dsDNA fragments ‘dilutes’ DNA-PK holoenzymes away from one another. Unlike isolated DSBs, where DNA-PK holoenzymes would normally
42 form ‘neatly’ on either side of the DSB breakpoint, at clustered breaks (surrounded by many small dsDNA fragments) the holoenzymes may not be assembled in opposition to one another and hence could be less able to trans-autophosphorylate. However, the literature is not entirely consistent, and one study using confocal microscopy indicates that DNA-PKcs foci are present within γH2AX/53BP1 foci in cells exposed to heavy ions, suggesting the enzyme is not absent from high-LET IR-induced lesions 139. Another study found that the chemical inhibition of DNA-PK by wortmannin had a smaller impact on α-particle-induced DSB repair versus low-LET IR157. The authors here interpreted their findings in support of the concept that clustered DNA damage may not be a suitable template for recombination-mediated rejoining, which has also been suggested elsewhere155,158. Wortmannin is an irreversible catalytic site inhibitor that also inhibits ATM, ATR and numerous other PIKK and PI3K enzymes. Possibly DNA-PK is already inhibited following HZE IR by fragments, and further inhibition would have a relatively smaller effect compared to low-LET IR. The absence or inhibition of DNA- PKcs may differentially influence the outcome of DSB repair, with greater use of HR observed without active DNA-PKcs and reduced HR seen when DNA-PKcs autophosphorylation is blocked 127102. This complicates the interpretation of previous studies into the role of DNA-PKcs in the repair of high-LET IR-induced damage, and more work is justified.
1.7 Radon Inhalation – The Most Common Form of Human IR Exposure Exposure to α particles via radon gas inhalation is by far the most prevalent source of IR exposure for 20th and 21st century humanity, accounting for approximately half of the world population’s annual exposure159. Indeed, radon inhalation is the greatest single source of lifetime radiation exposure for most modern humans160, correlating with increased rates of lung and hematologic malignant disease, melanoma, kidney cancers and certain childhood cancers159,161–163 and areas with higher residential measurement of radon appear to have higher lung cancer mortality rates164. Radioactive radon (222Rn) gas arises naturally from the decay of solid, earth-bound radium (226Ra), itself part of the decay series of uranium (238U). As the first and only gaseous element within that decay series, radon gas will diffuse upwards from the high-pressure environment of soil gases,
43 through foundational fissures and may concentrate within enclosed, well-insulated spaces such as households. As homes are heated and thermal stacking generates pressure differentials, and radon with other soil gases is actively drawn up through foundations to accumulate within indoor air. It is worth noting that this exposure modality was most likely uncommon during the evolution of homo sapiens and ancestral primates, as significant dose exposure would coincide only with our relatively recent adoption of increasingly air-sealed, completely encased shelters in direct contact with soils; this may have a bearing on the high LET IR induced DNA repair ‘problem’ outlined in previous sections. With a relatively short (3.8 day) half-life, radon gas decays rapidly into solid (and still radioactive) polonium (218Po), precipitating onto surfaces, dust particles or within the lungs. With a relatively short penetrance, α-particles do the most damage to cells immediately surrounding the location where the 222Rn initially precipitated, most often the mucosal lining of the lungs165. Radon decay within the lungs results in tissue α- particle bombardment and precipitation of solid polonium within lung mucosal linings; precipitated polonium (and other radon ‘daughters’) attached to household dust can also be inhaled165,166. Radon is typically measured and reported in Becquerels (Bq); 1 Bq is defined as the activity of a quantity of radioactive material in which 1 atomic nucleus decays per second – essentially the number of alpha particle emissions per second. Radon concentrations are expressed as Bq/m3 air. Radon decay products emit multiple α particles and β particles before becoming solid radioactive lead (210Pb). Alpha particles carry enough energy to remove electrons from other molecules, leading to ionization. DNA is easily ionized by α particles and breaks apart as α particles travel through tissue, generating difficult-to-repair clustered DNA damage as outlined in detail within previous sections. DNA damage leads to genetic mutation that increases cancer risk with each new α-particle emission. Evidence in animal studies show radon gas can dissolve in blood to be carried to distal bodily locations to adversely impact cell health167,168. Perhaps not surprisingly, there is convincing evidence that populations with a high natural background exposure to radon gas show adverse health effects164,169, with radon inhalation most clearly linked to lung cancer 170–172.
44 In 2004, a collaborative European analysis of 13 case control studies of residential radon exposure entailing more than 7,000 cases of lung cancer and 14,000 controls across 9 countries was published159. Darby et al. revealed a 16% relative risk increase in lung cancer for every 100 Bq/m3. The analysis supported a linear no threshold dose-response and estimated that radon of responsible for 2% of all cancer deaths in Europe. (Lung cancer in aggregate accounts for approx. 17.6% of the global cancer deaths173. Around the same time, a combined analysis was carried out in North America, consisting of seven studies with over 4,000 lung cancer cases and over 5,000 controls 171. In fact, Field et al. (2002) have shown that the power of an epidemiologic study to detect an excess risk from residential radon exposure is enhanced by linking spatially disparate radon concentrations with the subject’s retrospective mobility to obtain a true estimate of exposure171. For never-smokers with lung cancer, α-particle exposure via radon inhalation is considered the most common cause of disease174,175; as a separate category to tobacco-induced disease, never smoker lung cancer is the 7th leading cause of cancer death in the world and underlies an estimated 4-16% of all lung cancers depending on region175–177. Approximately 10–15% of all lung cancers occur in never-smokers, making never- smoker lung cancer a leading cause of mortality178,179. Some genetic factors proposed to predispose individuals to lung cancer involve DNA repair pathways, such as XRCC1 and ERCC2180. Future work will be required to identify genetic risk factors for high-LET IR induced lung cancer. As discussed earlier, radon gas exposure equates with chronic, low-dose α-particle irradiation. Four α-particles are emitted for every atom of 222Rn before it becomes non- radio isotopic lead, with three emissions occurring in the first day and the last taking several decades due to the long half-life (22.3 yr) of radioactive 210Pb. 222Rn-derived, α- particle-induced DSBs are likely slow-repairing due to their clustered nature and potentially the ability of high-LET IR induced DNA fragments to inhibit Ku and DNA- PKcs function (discussed above). Many governmental health authorities recommend maximum acceptable exposure limit or ‘action level’ of 100–200 Bq/m3, above which mitigation should be carried out to reduce household radon gas levels27,159,171,176. For this country, Health Canada indicates that 200 Bq/m3 represents the maximum acceptable exposure level before certain and serious health risks176. Anyone aged 65 years or less
45 who chronically inhales radon is at increased risk for lung cancer in their lifetime, with children and teens most affected159,172,181,182. In the Canadian province of Alberta, lung cancer is diagnosed in 2,150 people yearly, with mortality at 1610 deaths/year; of the 2150, 358 are never-smokers50,176. It is widely accepted that cancer prevention is preferable to the physical, emotional, social and economic cost of cancer diagnosis and therapy. Thus, radon exposure prevention represents an effective way to reduce cancer burden.
1.8 Other Sources of High LET IR Exposure The biological effects of heavy-ion irradiation (HZE) are an area of rapidly growing interest, since HZE particles are a major concern for manned deep space exploration. Although a relatively minor percentage of annual terrestrial IR exposure, the protection conferred by the Earth’s magnetosphere lessens beyond the confines of the lower atmosphere, such that even regular airline travel183,184 significantly increases exposure to cosmic radiation. Hence, space travel completely outside of the Earth’s magnetosphere will inherently involve higher levels of HZE particle irradiation, which is known to cause severe biological effects due to their high charge and mass. HZE particles range from energetic protons to iron nuclei and will penetrate most materials due to their extreme energy, which is thousands of mega-electron volts185. These energetic charged particles most often arise from solar flares or coronal mass ejections, a massive burst of solar wind arising from the sun186. Notably, most HZE particles will pass through current spacecraft shielding technology to emerge still highly radioactive and damaging to cells185. Apart from the carcinogenic effects of HZE radiation on adult somatic tissue, the impact of ineffectively shielded HZE particle irradiation on pre/post-natal developing cells could potentially lead to serious health problems in multi-generational deep space crews. Many space agencies currently limit total career IR exposure to 1 Sievert (Sv), roughly a 5.5% increase to lifetime cancer risk. NASA’s current guidelines aim to limit career exposure doses below a 3% risk of exposure-induced death, calculated to be 0.6–1 Sv for females or 0.8–1.2 Sv for males187. A study based on data collected during the Mars Rovers’ journey has been used to estimate probable human exposure for a Mars-bound flight188. For the expected
46 one-way, a 560 million km trip, an astronaut within a spacecraft would be exposed to 331 mSv, a third of lifetime allowable IR exposure188. A hypothetical return journey to Earth would double that dosage to 0.66 Sv, not including any IR exposure on the surface of Mars. Radiation incurred from the Earth-to-Mars commute would be comprised mostly (95%) by galactic cosmic rays (GCR) and 5% from Solar Energetic Particles (SEP). GCR are typically composed of 85% high energy protons, 14% helium nuclei (i.e. α-particles) and 1% HZE particlesam188. Innovation in radiation shielding technology, as well as an increased understanding of genetic risk and resistance factors to high-LET IR-induced genome instability, will be required before humanity can safely engage in long-distance space travel. The use of heavy ions as a therapeutic agent in the treatment of cancer is growing in popularity, due to several attractive characteristics. Firstly, HZE particles have a high initial energy deposition that drops off sharply, meaning a targeted tumour mass receives the maximum dose with minimal collateral damage to healthy surrounding tissue189. Secondly, HZE particle irradiation is less subject to cell cycle variation or oxygenation, such that it will kill even oxygen-starved or non-replicating (and potentially tumour- initiating stem-like) cancer cells with much greater efficiency than low-LET IR189,190. Further, studies have indicated that HZE particle irradiation can induce apoptosis regardless of p53 status; thus, it could be possible to exploit high-LET IR in tumours where p53-dependent apoptosis is defective, overcoming cell-death resistance191. For a comprehensive review on the use on heavy-ion cancer therapy see189,192. Briefly though, given what we understand of the repair of high-LET IR-induced DSBs (discussed earlier), HZE particle irradiation may also prove to be particularly effective against HR-defective tumour groups, such as BRCA1/BRCA2-negative breast cancers, which would be substantially less able to resolve such damage. If technologically possible, inhibition of Artemis endonuclease activity may also prove to be useful to sensitize cells to the killing effects of HZE irradiation. A barrier to HZE therapy is the huge expense and size of the particle accelerator needed to generate heavy-ion beams; hence, this mode of treatment is currently highly limited in availability.
47 1.10 Hypothesis and Specific Aims
IR exposure poses a serious risk to health, can cause significant damage to DNA, and can result in cancer. Human beings are exposed to IR on a regular basis, and this contributes significantly to the cancer burden of the population. Radon gas is the single greatest source of lifetime IR exposure, and this alone has been estimated to cause 2% of cancer in a European cohort159. Within this exposure modality, residential radon gas exposure is likely the largest exposure route for most people. Thus, humans are most commonly exposed to high LET IR, which has different effects on cells than low LET radiation but, as it is difficult to study, our understanding of the radiobiology of this is lacking and an unmet need with important societal implications.
I HYPOTHESIZE that there are geographic and environmental design attributes that contribute to higher indoor radon concentrations, and thus influence the most common modality of human high LET IR exposure. I also suggest that an increased understanding of the genetic, behavioral and environmental risk factors of radon exposure will better position science, medicine and public health agencies to identify at risk populations, better serve existing radon-induced lung cancer patients and prevent future cases. However, to achieve this it must made possible to study alpha particle radiation in a much more robust, high throughput, cost-effective and generally ‘achievable’ manner than current methods allow. To investigate these hypotheses and address current technological limitations, I have formulated two larger aims, with multiple sub-aims:
Aim 1: Defining the geographic and environmental design attributes that contribute to high LET IR (indoor air radon gas) exposure within residential properties. a. Design and implement a multi-year study of residential radon gas exposure across Alberta, which is scalable across the Canadian Prairies. b. Use geospatial data to map the radon levels determine if any regions are of particular risk. c. Use self-reporting data sets to assess how environmental design metrics may influence indoor radon concentrations.
48 d. Survey occupant behaviors to isolate what (if any) factors influence residential indoor radon levels and ultimately human IR exposure. e. Compare and contrast different radon testing methods and quantitatively assesses their efficacy for informing domestic indoor air radon concentrations accurately and precisely.
Aim 2: Build a higher throughput methodology to study alpha particle IR-induced DNA damage repair within in a human cell line-based model system. a. Design, prototype, and build a method to irradiate large numbers of human cells with alpha particle radiation in such a way that is compatible with a broad spectrum of established DNA damage response and repair assays. b. Accurately determine the IR dosimetry of the irradiator using both medical physics and radiobiology approaches. c. Validate technology utility by investigating how DSB response signaling from the high LET irradiator contrasts with low LET irradiation under controlled conditions. d. Practically assess the utility of the technology for screening programs, specifically assessing compatibility with IRIF enumeration, alkaline COMET assay and cell viability methods in the presence/absence of small molecules or genetic mutants. e. Explore the compatibility of high LET IR irradiator technology with Saccharomyces cerevisiae genetic model systems of radiation sensitivity.
49 Chapter Two: Comprehensive Survey of Household Radon gas Levels and Risk Factors in Southern Alberta
2.1 Preface
This chapter represents peer-reviewed original research published in the Canadian Medical Association Journal (Open)193. It details an investigation of residential radon exposure in the Southern Alberta region of Western Canada and was the initial ‘pilot project’ that formed the basis of the expanded work described in Chapter 3. Some sections have been abridged from the original manuscript to avoid reiteration and supply added context. Referencing, formatting and numbering is also adapted for consistency. The unabridged manuscript is available online as a free open access article193.
This work served as the foundational basis for Bill 209, The Alberta Radon Awareness and Testing Act, authored by Member of Legislative Assembly Robyn Luff, and that was passed after three readings in the Alberta Legislature in December of 2017194. Ms. Luff acknowledges the contribution of my published work to Bill 209 in this article (Appendix G).
50 2.2 Abstract
The inhalation of naturally-occurring radon (222Rn) gas from indoor air exposes lung tissue to alpha particle bombardment, a highly mutagenic form of ionizing radiation that damages DNA and increases the lifetime risk of lung cancer. Our aim was to analyze household radon concentrations and risk factors in Southern Alberta, including Calgary, the 3rd largest Canadian metropolis. Methods: 2,382 residential homes from an area encompassing 82.5% of the Southern Alberta population were radon tested for 90+ days, per national guidelines. Subsequently remediated homes were retested to determine efficacy of radon reduction techniques in this region. Results: Homes had an average indoor air reading of 126 Bq/m3 radon, equating to an effective absorbed radiation dose of 3.2mSv/yr. 48% of households were ≥100 Bq/m3 and 12.4% were ≥200 Bq/m3, with homes measuring as low as <15 Bq/m3 and as high as 3,441Bq/m3. Maximum observed radon concentrations were 3-fold higher in properties built in the past 25 years compared to older homes, suggesting that modern building practices are increasing indoor air radon accumulation. Of 90 homes averaging 575 Bq/m3 before mitigation, radon suppression successfully reduced levels to an average of 32.5 Bq/m3. Interpretation: This work demonstrates that radon is a genuine public health concern in Southern Alberta, legitimatizes efforts to understand the consequences of radon exposure to the public, and suggest that radon testing and mitigation is likely to be an impactful cancer prevention strategy.
51 2.3 Introduction
Radon (222Rn) gas arises from the radioactive decay of radium, thorium and uranium- bearing soils and bedrock, and is prevalent in the North American Prairies. Radon permeates through soil under high pressure towards low or negative pressurized areas such as basements. As homes are heated and thermal stacking generates pressure differentials, radon is actively drawn up through foundations to accumulate within indoor air. Radon inhalation is the greatest source of lifetime radiation exposure 160, correlating with increased rates of myeloid leukemia, melanoma, kidney cancers, certain childhood cancers 161,162 and is estimated to be responsible for 2% of all cancer deaths 159. 222Rn has a 3.8-day half-life, with 50% decaying in that time to radioactive polonium and emitting alpha (α) particle radiation, which is, dose-for-dose, substantially more dangerous to health than x- or α-rays 163,195. Radon decay within lungs leads to tissue α-particle bombardment and precipitation of solid polonium within lung mucosal linings; precipitated polonium attached to household dust can also be inhaled 162. Radon decay products emit 3x α - and 2x β-particles before becoming solid 210Pb. Alpha particles carry enough energy to remove electrons from other molecules, leading to ionization. DNA is easily ionized by α -particles, and breaks apart as they travel through tissue, generating difficult-to-repair DNA damage that has a significantly higher dose effect than γ-rays 163,195,196. DNA damage leads to genetic mutation that increases cancer risks with each new α -particle emission197. Radiation is measured in Becquerels (Bq) that equal one radioactive decay event per second. Radon concentrations are expressed as Bq/m3 air, with 100 Bq/m3 increasing lifetime lung cancer risk by 16%8. Health Canada indicates that 200 Bq/m3 represents the maximum acceptable exposure before certain and serious health risks. Anyone chronically inhaling radon between age 0-65 is at increased risk of lung cancer in their lifetime, with children and teens most impacted 27,161,176,197. Globally, 25% of lung cancer patients are non-smokers, with most cases (in developed countries) caused directly by radon inhalation in homes and workplaces 181,182,198. 2,150 Albertans are diagnosed with lung cancer yearly, with mortality at 1,610 deaths/year; of these cases, 358 are never- smokers 176,199. Cancer prevention is preferable to the physical, emotional, social and economic cost of cancer diagnosis and therapy, and radon prevention represents an
52 effective way to reduce cancer burden. Motivated by this, our objective was to measure household radon, correlate levels with home metrics and examine the effectiveness of remediation strategies in Southern Alberta.
2.4 Methods
2.4.1 Setting and Design
From 2013-2016, ~3,000 Southern Alberta residents purchased ‘alpha track’ radon detectors as part of a study entitled the “Citizen Scientist Radon Testing Project”. Residents purchased radon tests for $45 each, which were then distributed centrally by our study. Homeowners and renters were equally eligible. Public outreach was achieved through print, online and TV/radio media in an untargeted manner. Participants consented to semi-anonymously provide long term average radon readings and home metrics, with data associated only with postal region. The survey collected construction year, build type, foundation type, floor tested and room of deployment. Rigorous care was taken to educate participants in the correct test deployment methods through communication with Canadian National Radon Protection Program-certified professionals.
2.4.2 Data Collection
Eighty percent of tests (n = 2,382) were returned and eligible for analysis; remaining tests were either not deployed correctly during the appropriate testing window or, for a small minority, spoiled by homeowner error. Commercial buildings, apartment blocks and mobile homes were not considered. Radon tests were closed passive etched track detectors made from CR-39 plastic film inside antistatic holders (Radtrak2, Landauer Radon, Inc., Glenwood, IL, USA) enclosed in electrically conductive housing with filtered openings to permit gas diffusion, with a typical linear range of 15 to 25,000 Bq/m3. To be read, CR-39 films are etched in 5.5N NaOH at 70°C for 15.5 min and scored using TrackEtch® software at C-NRPP accredited Landauer laboratories (ISO17025 certified). Controls included duplicates to ensure device reproducibility, spiked positives (to ensure accuracy) and non-deployed negatives (controlling for transport and storage prior to analysis). Readings throughout this study are in Bq/m3
53 rounded to the nearest whole number. The survey region included the City of Calgary, Cochrane, Okotoks, Airdrie, Canmore, Bragg Creek, Chestermere, High River, De Winton, Redwood Meadows and surrounding rural ‘municipal districts’; 2,018 homes were within Calgary and 364 were in surrounding townships. The median test duration was 103 days and >99.5% were deployed from October-April. I also collected follow-up data from 90 x ≥200 Bq/m3 households that opted for mitigation, mostly involving sub- slab depressurization, with a minority utilizing radon-impermeable membrane installation.
2.4.3 Statistical Analysis
Statistical analysis was carried out using SPSS software; unreported data was excluded. In the General Linear Model (GLM), radon concentrations were transformed to LnX to meet model assumptions. Type III sum of square was used in the GLM because the unbalanced dataset (e.g. for home types) and construction year was considered as a confounding covariate and was controlled against. Significance at p<0.05. One-way ANOVAs were carried out to test radon levels among groups (home metrics/year of construction/FSA), with Bonferroni post-hoc testing.
54
2.5 Results
2.5.1 Household Radon level in the Calgary metropolitan are
Indoor air radon levels for each home varied considerably, ranging from <5 Bq/m3 to 3,441 Bq/m3, and varied across all forward sortation areas (fig 2-1); the average reading was 126 Bq/m3. While some regions averaged slightly higher, there were no areas with uniformly low household radon levels and all areas contained homes with radon well above the national action level (fig 2-2). Cochrane (261 Bq/m3) and Okotoks/High River (194 Bq/m3) displayed highest average radon readings. Calgary itself averaged 122
Figure 2-1- Indoor air radon concentrations by postal code district in the greater Calgary metropolitan area. 2,382 individual home readings for indoor air radon are grouped by the first three digits of Canadian postal code. Darker coloured circles indicate multiple overlapping radon readings. High radon concentrations are documented almost universally across the region. (1. Airdrie East, 2. Canmore, 3. Central Foothills, 4. Cochrane, 5. High River, 6. Kananaskis Improvement District, 7. Okotoks, 8. Redwood Meadows, 9. Chestermere, 10. Symons Valley.
55 Figure 2-2 - Average indoor air radon concentrations by subdivision of the greater Calgary metropolitan area The percentage of homes between 0-100 Bq/m3, 100-200 Bq/m3 or >200 Bq/m3 for the four quadrants of Calgary and the surrounding towns, including number of homes tested in each region. Max = maximum observed radon reading in area; Min = minimum observed radon reading in area. In all cases, homes well above the maximum acceptable limit (200 Bq/m3) for Canada are documented.
Bq/m3, ranging from 72-164 Bq/m3 within city subdivisions (fig 2-2). To validate geographical effects, I grouped readings by forward sortation area (i.e. the first three postcode digits). Upon omitting low reporting (n<10) areas, I was left with 34 area located in or adjacent to Calgary. There was a statistically significant difference in household radon differed between forward sortation areas, f(33,2045)=3.272, p<0.001.
56 Average radon readings in Cochrane were significantly different from 26 of 34 regions (p<0.05), and High River different from one area in the southeast of Calgary (p<0.05). No areas in within Calgary city limits differed in a statistically significant sense.
2.5.2 Radon by home feature metrics
Our home metrics survey allowed us to examine factors associated with radon levels (Table 2-1). Basements had significantly higher radon vs ground or 1st floors(p≤0.001); utility spaces were also higher compared to living spaces (p<0.001) (fig 2-3). I also performed GLM analysis to test for influence of multiple variables on indoor radon, f(150,1332)=2.226, p<0.001. To isolate the impact of home metrics, construction year
Figure 2-3 - Average indoor air radon concentrations by home features Box plots showing min/max spread and mean of indoor air radon tests within home descriptor groupings (*p ≤ 0.001, Bonferroni Post hoc testing on one-way ANOVAs).
57 Figure 2-4 - Maximum observed household air radon concentrations are significantly higher in homes built in the past 25 years. (A) Indoor air radon measured in homes built between 1992 and 2016 versus 1991 and earlier. (B) The reported square footage of homes constructed between 1945 and 2016. (C) Average home square footages across decades. was controlled against. Of the tested effects, only two were at or close to significant: [Build x Room x Area], [Build x Area], p<0.05 (Table 2-2).
2.5.3 Radon by home age and mitigation status
A strong correlation was found between home construction year and radon concentration. A Mann-Whitney U test was run to determine if there were differences radon levels between homes built in the last 25 years and prior. Median engagement score was statistically significantly higher new homes (85 Bq/m3) than in old (103 Bq/m3), U = 653,867.500, z = 6.203, p<0.001 (fig 2-4A). An unbiased survey of 1,632 area homes listed for sale in 2016 demonstrates that the home age distribution in our study is comparable to norms of the region (fig 2-4B). While doing this, I recorded home floorplan sizes, noting that these have doubled over the past 65 years in Southern Alberta. Homes built from 1992-2016 were an average 2,384.9 ft2 in size (n=914), 51% larger
58 Figure 2-5 - Pre- and post-mitigation household air radon concentrations for all homes initially measuring ≥200 Bq/m3. Red bars indicate radon levels prior to mitigation. Blue bars are the corresponding post-mitigation radon level for that same household. than homes built in the previous 90 years (1,578.9 ft2, n=725) (fig 2-4C). Of 90 homes averaging 575 Bq/m3 before mitigation, radon suppression successfully reduced levels to an average of 32.5 Bq/m3, with the most striking case being a home at 3,441 Bq/m3 successfully reduced by 97.5% to 86 Bq/m3 (fig 2-5).
2.6 Interpretation
2.6.1 Main Findings
Our principle findings are that 48% of Southern Alberta homes tested were >100 Bq/m3 and 12.4% exceeded the 200 Bq/m3 Health Canada guideline. Homes built in the past 25 years had, on average, 31.5% higher radon levels compared to older households. I interpret the slightly higher radon levels observed in basements and utility spaces to be attributable to reduced ventilation and proximity to radon entry points. While there are some other apparent minor trends, I am reluctant to further interpret these in terms of
59 radon prediction without more balanced datasets. Homes remediated for high radon demonstrate that mitigation is effective at reducing levels <100 Bq/m3, with 100% success in this group and typically achieving 92% radon reduction. Thus, I conclude that radon mitigation by C-NRPP-certified workers is generally effective in Southern Alberta.
2.6.2 Explanation and Comparison
Our data indicates a strong correlation between construction year and indoor air radon level, while geographical location (within this region) was not an effective predictor. These findings suggest that millennial home engineering practices are creating environments that accumulate radon in greater indoor air concentrations. I speculate that this could be due to: Energy-efficient home insulation practices that reduce heat-loss but also often suppress air exchange. Indeed, increasing air-tightness can elevate mean radon concentrations by 56.6%200. This effect also makes radon a more pressing concern in colder climate countries such as Canada. Home floor plan sizes in Alberta have steadily increased over time (fig 2-4B). Since concrete contracts as it cures in a fixed ratio with the size of poured slab201,202, larger floors are subject to greater shrinkage and, consequently, larger floor-to-foundation gaps enabling more radon entry. This is likely exacerbated by the fact that concrete shrinkage has increased, reportedly due to a scarcity of good quality aggregates and subsequent use of recycled concrete with mineral additives (such as fly ash)202203. (iii) Building height has also increased over time, with 2-3 story homes with vaulted ceilings becoming the norm. Loftier homes, like taller chimneys, exhibit potent thermal stack effects that generate powerful negative pressures at basement level that draw up ever more soil gas (radon) into the indoor air.
I found radon levels higher than those previously reported in The Cross-Canada Survey of Radon Concentrations in Homes176, which showed 8.1% of 86 Calgary Health Region homes were ≥200 Bq/m3, with none ≥600 Bq/m3. Our new data, based on 2,382 readings in the same region, reveals that 12.4% of homes are between 200 and 3441 Bq/m3. This
60 indicates that Southern Albertan homes contain higher radon than Canada as a whole (6.9% ≥200 Bq/m3), in line with ≥200 Bq/m3 radon home estimates for Manitoba (19%) and Saskatchewan (9%)176.
2.6.3 Strengths and limitations
- This work represents one of the largest municipal studies of household radon in Canadian history and has analyzed radon not only by region (the historic norm for previous work176) but correlates it with specific Canadian-built home metrics. While there may be interactions between build type and location, I conclude that such a correlation is only weakly demonstrated and more balanced data (e.g. single family detached homes represented >90% of build type in our study) is needed to test this further. For some home metric data points, we fell short of 100% response rate, with an average 89% of information returned. I note that our survey was a ‘convenience sample’ achieved through local public outreach, while the Health Canada survey176 was a ‘random sample’ contacted through phone survey; both datasets were, however, collected using the same guidelines. Our sampling through public outreach potentially applied a selection bias, although our outreach was untargeted, and we accepted all valid data from Southern Albertan residencies, potentially minimizing this impact. Costs incurred by the participants may have also biased our sample group, although, based on figure 2-4b, this did not bias the distribution of home ages within our study compared to the norm for the region. Our work focused entirely on non-commercial residences and in future it will be important to expand this analysis to encompass daycares, workplaces, etc. Based on census information, City of Calgary residences are 55.8% single family homes, 6.6% duplexes, 22.6% condo/apartments, 10.6% townhouses and 4.4% ‘other’. Our data set is comprised of 92% single family detached homes, 5% duplexes, 0.5% condo/apartments and 2.5% townhouses, indicating an (intended) under-representation of apartment blocks in favor of detached homes. Based on occupancy, more than 5X the regional population live in single family detached homes compared to Condo/Apartments. As such, our sample covers the most at-risk and largest section of the population of interest and
61 represents the first major systematic examination of radon with community-level resolution for a major Canadian city.
2.6.4 Implications
The 2015 population of our study region was 1.43 million, encompassing 82.5% of the Southern Alberta population. Census data estimates this area to contain 463,682 single family residences and, so, 12.4% homes >200 Bq/m3 equates with ~57,500 homes or, considering occupants per household (2.6), ~150,000 residents - a substantial population whose lifetime risk of lung cancer is avoidably increased. Our metrics also suggest that basement-suite occupants (common practice in the region) are exposed to greater radon. Our data also indicates that populations at risk from radon (12.4%) are comparable to those at risk of tobacco-related cancer, as smoking rates in Alberta are 19%204. Since exceptionally high radon readings were observed across the region, I conclude that there are no reliable ‘safe’ areas and all neighborhoods are potentially at risk. I speculate that remediation in Southern Alberta maybe so effective (100% observed success, fig 2-5) due to the porous gravel placed beneath most region homes, enabling soil gas flow across foundations to mitigation devices. Radon is an avoidable cause of cancer with serious economic consequences to health care systems, individuals and society. 2,150 lung cancer cases are diagnosed in Alberta annually and ~358 are in never-smokers and are thought to be attributed to radon. Direct costs paid by the Alberta healthcare system for one case of lung cancer are $24,055/person21. Studies indicate that indirect costs, encompassing lost productivity, informal care-giving and costs borne outside the healthcare system, are as much as direct costs22. Factoring this, total Alberta costs of radon-induced lung cancer are estimated to be at least $17 million/yr.
2.7 Conclusion
Southern Alberta is a region of high geologic radon potential and home to ~1.4 million people, including Calgary, the 3rd largest Canadian census district 27. This study shows that radon – an established carcinogen – is of genuine concern in this region and
62 legitimatizes efforts to understand the consequences of exposure to public health. Approximately 2,000 Albertans are diagnosed with lung cancer annually and 25% are never-smokers 12,28. I demonstrate that 48% of area homes contain >100 Bq/m3 indoor air radon; 12.4% exceeded the 200 Bq/m3 maximum acceptable limit for Canada. Homes built in the past 25 years have significantly higher radon compared to older homes, and remediation methods are effective at reducing exposure, suggesting that radon testing and mitigation will be an impactful cancer prevention strategy.
63 Table 2-1- Radon levels by home metrics
Metric No. of homes mean radon± SEM, Observed radon level (n = 2382) Bq/m3 Min/Max Bq/m3 Year of construction 1890-1939 60 72 ± 7 < 15 361 1940-1959 174 94 ± 7 < 15 722 1960-1979 547 120 ± 4 < 15 1100 1980-1999 721 106 ± 3 < 15 1069 2000-2009 432 132 ± 6 < 15 1274 2010 or later 195 218 ± 29 < 15 3441 Not reported 253 - Building type Condominium/apartment 12 81 ± 17 < 15 167 Single-family detached 1819 126 ± 4 < 15 3441 Duplex 113 108 ± 10 < 15 925 Townhouse 50 76 ± 12 < 15 498 Not reported 388 136 ± 6 < 15 1014 Foundation Basement 1973 125 ± 4 < 15 3441 Slab 28 120 ± 21 < 15 472 Bilevel 89 106 ± 6 < 15 309 Crawl space 30 95 ± 20 < 15 607 Not reported 262 144 ± 8 < 15 1014 Floor tested Basement 1360 137 ± 4 < 15 3227 Bilevel 58 91 ± 6 < 15 240 Ground 485 62 ± 4 < 15 1005 First 246 101 ± 7 < 15 1274 Second/third 31 92 ± 13 17 326 Not reported 202 176 ± 22 17 3441 Room Bedroom 282 125 ± 7 < 15 1198 Living space 1740 117 ± 3 < 15 2346 Utility space 105 224 ± 48 < 15 3441 Not reported 255 147 ± 8 < 15 1014 Total 2382 127 ± 3 < 15 3441
Table 2-1 - Radon levels by home metrics. 2,382 household air radon readings grouped according to reported building metrics with average indoor air radon concentrations, error of the mean, and range. Note: SEM = standard error of the mean.
64
Table 2-2 - Linear Model of Radon Levels Based on Home Metrics Variable df Sum of Mean F p value square square Corrected model 150 120.519‡ 0.803 2.226 0.000 Intercept 1 6.857 6.857 19.00 0.000 1 Year of construction† 1 18.433 18.433 51.08 0.000 0 Found 3 2.150 0.717 1.986 0.114 Build 3 0.775 0.258 0.716 0.542 Floor 4 2.166 0.541 1.500 0.200 Room 2 1.996 0.998 2.765 0.063 Area 4 2.593 0.648 1.797 0.127 Found × build 2 1.776 0.888 2.461 0.086 Found × floor 9 3.891 0.432 1.198 0.292 Found × room 3 0.066 0.022 0.061 0.980 Found × area 11 3.992 0.363 1.006 0.439 Build × floor 7 3.288 0.470 1.302 0.246 Build × room 5 1.054 0.211 0.584 0.712 Build × area 10 6.634 0.663 1.838 0.050 Floor × room 7 5.929 0.847 2.347 0.022 Floor × area 16 5.395 0.337 0.934 0.529 Room × area 8 5.514 0.689 1.910 0.055 Found × build × floor 0 0.000 - - - Found × build × room 0 0.000 - - - Found × build × area 0 0.000 - - - Found × floor × room 2 1.927 0.964 2.670 0.070 Found × floor × area 11 2.707 0.246 0.682 0.757 Found × room × area 2 0.855 0.427 1.184 0.306 Build × floor × room 1 0.124 0.124 0.343 0.558 Build × floor × area 10 4.935 0.494 1.368 0.190 Build × room × area 6 4.581 0.764 2.116 0.049 Floor × room × area 12 3.283 0.274 0.758 0.694 Found × build × floor × room 0 0.000 - - - Found × build × floor × area 0 0.000 - - -
65 Found × build × room × area 0 0.000 - - - Found × floor × room × area 0 0.000 - - - Build × floor × room × area 1 0.074 0.074 0.206 0.650 Found × build × floor × room × area 0 0.000 - - -
Error 1332 480.679 0.361 Total 1483 31 569.6 83 Corrected total 1482 601.199
Table 2-2 - Linear Model of Radon Levels Based on Home Metrics effect estimates, and descriptive statistics based on sum of squares type III general linear model where year of construction is controlled against. Notes: aSS= sum of square, bMS = mean square, cDependent Variable: Ln(Radon Concentration), dCovariate: Year of Construction, eFound = foundation type, fBuild = build type, gFloor = floor tested, hRoom = room tested, build = build type, df = degrees of freedom, floor = floor tested, found = foundation type, room = room tested. *Dependent variable ln(radon concentration). †Covariate. ‡R2 = 0.20, adjusted R2 = 0.11.
66 Chapter Three: Environmental design and behavioral variables influencing radon gas exposure and testing accuracy in the Canadian Prairies
3.1 Introduction
Well established and effective testing and mitigation techniques exist for residential radon; thus, radon exposure represents a readily preventable cause of the most lethal and common cancer. Radon gas arises geologically from decaying radioactive radium in soils and is prevalent across Earth. Although arising naturally, radon can be concentrated to very unnatural levels within human-made buildings, typically outdoor concentration is 10 Bq/m3 whereas indoor levels can be orders or magnitudes higher than this, previously 7% of Canadian homes tested were over 200 Bq/m3 176. Decaying 222Rn emits alpha particle radiation, which triggers a chronic inflammatory process by severely damaging DNA in such a way that is almost impossible for our bodies to repair without introducing genetic errors 3. Such errors trigger “genomic instability”, a self-propagating cycle of DNA alteration that drives cancer formation. Hence, radon is listed as a category 1 carcinogen, meaning that it is absolutely known to cause cancer in humans. International Agency for Research on Cancer (IARC) categorises radon as a Group 1 agent meaning there is evidence of carcinogenic action with human exposure205. Canada contains many radon-generating regions, and Canadians have constructed population centres across almost all of them, although this does not necessitate that all buildings contain high radon4,5. Indeed, there are three factors needed to incur hazardous radon exposure: (i) a geologic source and pathway (upwards) for radon, (ii) environmental design metrics that actively draw up and concentrate radon and (iii) human behaviour that prolongs exposure or increases radon concentrations. These latter two variables are modifiable and are of interest in terms of cancer prevention. As awareness regarding radon’s health hazards increases, stakeholders are asking how to establish best practice for calculating an individual’s exposure. From a medical practice and research perspective, establishing historic and ongoing radon exposure is important cancer risk information, like smoking history. This can enable earlier cancer diagnosis, referral to cancer screening programs and large longitudinal health studies. Government and regulatory bodies are also determined to define best practice for radon testing within
67 economic and social scenarios such as licensing business (such as daycares), real estate transactions, home inspections and building renovation. Thus, it is important to ask: (i) what is the ideal radon testing method to establish radon exposure data reliable for high stakes decision-making, and (2) whether different radon testing methods are all equivalent in distinct environmental design, regional and/or seasonal contexts. Radiation is measured in Becquerels (Bq) that represents one radioactive decay event per second. A 16% increase in relative risk of lung cancer is measurable per ≥100 Becquerel/m3 (Bq/m3) chronic radon inhalation27,159. In Canada, 200 Bq/m3 is considered maximum allowable by Health Canada, with advice being to strive for as low as reasonably achievable. In our 2017 cross-sectional study of 2,385 Southern Alberta homes, I found that 1 in 8 contained ≥200 Bq/m3and that newer homes contained higher radon versus older properties193. It was unclear, however, whether this applied more broadly across Canada, what specific environmental design metric(s) and/or behaviour(s) contributed to this effect and which, if any, were modifiable. Motivated by this, I measured household radon across the largest Canadian Prairie provinces, Alberta and Saskatchewan (total area = 1,313,748 km2, approximately the same land area as the Republic of South Africa), and coupled this to geospatial analysis, an interrogation of how environmental design metrics and associated behaviours correlate with radon levels and, within a subset of regional homes, multiple modalities of radon testing.
3.2 Methods
3.2.1 Main Study Setting and Design
From 2010-2018, 10,731 Alberta and Saskatchewan residents purchased or were provided alpha track 90+ day radon detectors that they then deployed and returned for analysis, receiving their specific radon reading confidentially. The majority (8,033) of these citizen scientist participants were enrolled via the ‘Evict Radon’ study, which additionally surveyed home metrics and behavioural information. 2,698 tests were generously provided by the Lung Associations of Alberta/NWT and Saskatchewan, coupled to geospatial information (Forward Sortation Area, FSA) and testing period information. 515 tests were obtained from the Saskatchewan Health Authority. Participants obtained
68 tests at cost for $45-60 each (depending on year, as currency exchange fluctuations in later years increased costs), which were then distributed centrally by researchers. Homeowners and renters were equally eligible. Public outreach was achieved through print, online and TV/radio media in an untargeted manner. All participants consented to semi-anonymously provide researchers radon and/or home metric and behavioural data, with the understanding they would never be publicly identified. Participants were permitted to withdraw at any time. The Evict Radon cohort survey collected construction year, build type, foundation type, floor tested, room of deployment, ceiling heights, thermostat settings, window opening behaviour, basement and ground floor surface area (square footage) and thermostat settings. Rigorous care was taken to educate participants in the correct test deployment methods through communication with Canadian National Radon Protection Program-certified professionals and close adherence of advice and testing protocols to Health Canadas guidelines.
3.2.2 Radon Test Comparison Subsets
The 2017-2018 Evict Radon cohort was segregated into representative groups encompassing a normalized distribution for build year and location within the region, and then 1000 homes were randomly selected for additional deployment of an alpha track 5- day radon detector to be placed side-by-side with the 90+ day radon test for the final 5- days of the test period. Of those homes who deployed 5-day alpha track tests during winter months (March 2018), 100 were invited to deploy a second 5-day alpha track test in the same location during the summer (July-August 2018). A subset of 30 homes, confined within a specific geographic region to control for meteorology trends (NW quadrant of the city of Calgary), were also selected for real time continuous radon monitoring, coupled with simultaneous 5-day electret and alpha track test device deployment.
3.2.3 Data Collection
A total of 11,727 (fig 3-4a) radon tests were returned and eligible for analysis; although since this work represents the aggregation of multiple testing efforts across several years
69 the accompany data varies from the radon reading, FSA and test date to a fully complete evict radon survey (sup.1). Commercial buildings, apartment blocks and mobile homes were not considered. Alpha track radon tests were closed passive etched track detectors made from CR-39 plastic film inside antistatic and electrically conductive housing with filtered openings to permit gas diffusion, with a typical linear range of 15 to 25,000 Bq/m3. To be read, CR-39 films are etched in 5.5N NaOH at 70°C for 15.5 min and scored using TrackEtch® software at C-NRPP accredited Landauer laboratories (ISO17025 certified). 90+ day alpha track tests were Radtrak2 devices, whilst 5-day alpha track tests were Duotrack devices both obtained from and analyzed by Radonova (USA and EU). Electret devices were e-PERM short term radon tests from and analyzed by AGAT laboratories (Canada). Continuous radon monitors were RadonEye+ ion chamber devices from FTLab (South Korea). Controls included duplicates to ensure device reproducibility, spiked positives (to ensure accuracy) and non-deployed negatives (controlling for transport and storage prior to analysis). Readings throughout this study are in Bq/m3 rounded to the nearest whole number. The survey region included all of Alberta and Saskatchewan. For long term alpha track test devices, the median test duration was 103 days and >91% were deployed from October-April.
3.2.4 Geospatial Analysis
All readings with accompanying geospatial data (n=11402) were mapped to federal electoral districts (ED) encompassing administrative divisions, as defined by the 2013 Representation Order, with approximately equal populations within provinces (based on the 2011 census, Ab=107,000, Sk=74,000) (Elections Canada, 2015)(fig 3-1). Geospatial analysis was run in ArcGIS pro, maps were produced with ArcGIS pro or Inkscape. The most precise division I report is FSA which were also mapped when possible. I also report unique regions comprised of consolidated EDs for ease of interpretation, typically these are the joining of multiple districts from the same city (Table 3). It should be noted that geotagging relies on accurate reporting of postal code, and several tests can not me matched with certainty to an ED (n=325).
70 3.2.5 Statistical Analysis
Statistical analysis was carried out using R (3.5.1) and RStudio (1.1.456); unreported or uninterpretable data were excluded. One-way ANOVAs were carried out to test radon levels between groups (home metrics/year of construction/ED), with Bonferroni-Holm post-hoc testing carried out for pairwise comparisons if the ANOVA reached significance. Regressions were run on Ln transformed radon concentrations. Stepwise subset regression analysis was run using an exhaustive search with all predictors (dummied as necessary). Minimised Bayesian Information Criterion (BIC) was used for model selection. AIC, r-squared and adjusted r-squared were also calculated for the exhaustive subset search.
3.3 Results
3.3.1 Extent, Structure, and Overall statistics of the Study
The reported datasets are all tests for the concentrations of radon gas in indoor air of residential buildings. All numbers are reported in Bq/m3 rounded to the nearest whole number. For long-term (90+ day) radon tests, I present data from tests carried out according to Health Canada’s recommendations, specifically that one long-term detector is placed on the lowest level of the home that is occupied for more than 4 hours per day, for a minimum of 3 months during the typical Canadian heating season (October to April), with CARST-certified devices. The survey area is comprised of some 11,726 residential radon tests, of those tests 55% were above 100 Bq/m3 (n = 6,257) and 18% above 200 Bq/m3 (n= 2,086), the arithmetic mean +/- SD for all tests was 146 Bq/m3 +/- 172 (fig 3-4a).
The survey encompassed a wide variety of home build types, as reported in the core survey data set. Analysis of variance showed significant differences between these build types (F (7, 5011 = 5.15, p <0 .0005)) (fig3-2a), and post hoc analysis is dominated by the semi-detached townhouse group which skewed significantly lower compared to multiple other home types (table 3-1). Radon readings were statistically higher when the test was placed in the basement level (F (4, 5063 = 8.20, p < 0.0001)), as compared to the ground
71 and/or first floors; however, there were no apparent differences in tests carried out in different room types (F (3, 5046 = 1.67, p < 0.17)). Testing was principally carried out in the Canadian Provinces of Alberta (AB) and Saskatchewan (SK) (fig 3-1). Using the electoral division (ED) to visualize how well my testing sample represents the regional population, I found testing encompassed most Alberta (32 of 34) and Saskatchewan EDs (13 of 14), which all had a minimum of 30 reported tests and an average of 237 test results per ED (range 14-1,016 tests per ED) (fig 3-1, table 3-2). There was a total of 21 tests reporting household radon level in homes outside AB and SK (tab 1). The most precise division I report is FSA, which were also geographically mapped when possible. I obtained radon readings from a total of 199 FSAs, albeit 120 FSA regions reported less than 30 tests (range n = 30-688, radon = 81- 362 Bq/m3).
3.3.2 Geographical Analysis of the Alberta and Saskatchewan Radon Testing
At a provincial level, the mean radon level for SK homes was 218 Bq/m3 (min <15 Bq/m3, max 2985 Bq/m3), an average reading that was significantly (p<0.0001) higher (by 88 Bq/m3) than the average radon level in the neighboring province of AB, which was 130 Bq/m3 (min <15 Bq/m3, max 7199 Bq/m3)(fig 3-3). There were modest differences observed within AB, with the Southern part of the province trending higher (154 Bq/m3) than Central and Northern areas (123 Bq/m3). My dataset is broadly weighted in a manner that reflects the relative population densities of AB versus SK, with 83.5% (9,507) of radon readings from AB households and 16.5% (1,874) of readings from SK homes. The total population for both provinces is approximately 5.4 million people, with ~80% of the population in AB (4,307,110 residents reported by the Alberta government as of July 1st, 2018), and ~20% in SK (1,098,352 residents reported by the Saskatchewan government
72 Figure 3-1 – Extent of Residential Radon testing in Alberta and Saskatchewan. Cartogram showing the level of response in the AB and SK provinces according to ED divisions. ED’s contain roughly similar populations. All but three ED’s had more than 30 tests, the majority had more than 100 (min-max: 14-1016). See table for further details.
as of May 10th, 2016). Thus, whilst my data is very slightly weighted in favour of AB, this is not substantial and average radon readings encompass a reasonable population- weighted distribution (fig 3-3). Consolidating some of the EDs for ease of interpretation (fig 3-4b), I see that the average radon level in all regions is above 100 Bq/m3, with the overall average for these two provinces together being 146 Bq/m3 (min <15 Bq/m3, max 7199 Bq/m3). Looking specifically at the percentage of homes testing above the 200 Bq/m3 Health Canada action
73 Figure 3-2 – Residential radon testing results divided by testing location with the residence. (a-c) Spread of residential radon testing and mean radon level (+/- 95% CI) across type of construction, room, and floor of testing. Pairwise post hoc testing details in Table 9. (a) build type (p<0.0005) (b) floor of testing (p<0.0005) (c) room of testing. (Not significant)
level, 18% of homes tested above this in AB, while 35% tested above this level in SK (fig 3-4b). If we focus on the major Prairie province urban centres, Edmonton (pop. 1.32
74 million), Saskatoon (pop. 0.3 million) and Calgary (pop. 1.39 million) all had a comparable percentage of homes with radon at or exceeding 200 Bq/m3, 17%, 16% and 17% respectively, and a comparable mean radon level (121-132 Bq/m3). By contrast, 50% of homes in the City of Regina (pop. 0.24 million) were at or exceeded 200 Bq/m3, with an arithmetic mean radon level of 296 Bq/m3 (fig 3-3, table 3-1). At the individual ED level, the arithmetic mean radon levels are statistically different significant (p<0.0005) (Appendix F), and some 133 pairwise post hoc Holms-Bonferroni corrected Tukey tests
Figure 3-3 – Map showing radon testing statistics across Alberta and Saskatchewan. Map showing radon testing statistics of across the major testing region. Statistics report for all reporting, each province. Map inserts showing radon testing statistics for the major cities in the region, with rural/small town levels of geographic division. All divisions based on ED boundaries. (Arithmetic) Mean radon levels reported in centre of pie chart showing percentages within the indicated levels.
75 Figure 3-4 – Full spread of all residential radon tests, also divided by geographic region (a) Every completed test charted on a ln axis (n=11,726). 18% over ≥200 Bq/m3, (mean +/- 95% CI) (Min-max: <15-7199 Bq/m3) (b) All mappable tests charted on a log axis according to ED based population centre division. (S. AB= southern Alberta, N.&C. AB = North and Central Alberta, all greater regions that encompass another region exclude the encompassed regions, e.g. Saskatchewan column contains no Saskatoon or Regina data, see figure 11 for boundaries) (n=11402) (yellow; 0-99 Bq/m3, orange; 100-199 Bq/m3, red; ≥200 Bq/m3
resulted in p values less than 0.0005; 66 of these pair tests included one of the two Regina ED’s. The most precise level of geographic designation utilized within my study is the
76 forward sortation area (FSA) code. There does appear to be a significant difference in the means for a household radon concentration in certain regions (F (198, 11203) = 7.05, p < 0.0001), and pairwise post hoc testing with Tukey plus Holm-Bonferroni correction resulted in 89 pairwise mean comparisons with p<0.005. Once again, most (74) of these matches contained 1 of the 3 Regina districts. The average radon concentration in the Regina EDs (47007, 47009) were 309 Bq/m3, 282 Bq/m3 respectively, compared to the overall mean of 146 Bq/m3 from this dataset. Collectively, this highlights the City of Regina as a hotspot for household radon within the survey region, with other major urban areas being broadly comparable with one another and almost half the mean radon level seen in Regina. However, whilst Regina may have contained the highest average radon, the highest maximum household radon reading (7,199 Bq/m3) was observed within Southern Alberta. Radon data for a complete list of all municipalities with >30 readings is described in table 3-2 and appendix F. All areas examined to date contained homes exceeding maximum exposure thresholds.
3.3.3 Household Radon Concentrations as a Function of Year of Building Construction
In my initial Southern Albertan radon study, I provided evidence of increased probability of higher radon levels in newer homes. In the updated analysis, a regression analysis predicts an increase of indoor radon concentration with each additional year, in the year of construction scale (F (1, 7791 = 235.2, p < .0001)), although this model only explains a small fraction of the overall variance in radon concentration (adj. r2 = 0.03) (fig 3-5b). The trend is clearer when looking at evenly distributed (quantile) groups of homes (i.e. with each quantile containing an even number of readings) separated by build year periods (1800-1974, 1975-1993, 1994-2004 and 2005-2018) (table 3-3), In all cases, both the mean radon level and the percentage of homes exceeding 200 Bq/m3 maximum exposure threshold increases with time (that is, the “newness” of homes). Indeed, the number of properties >200 Bq/m3 almost doubles between the first and last quantile, whilst mean radon increases by >50 Bq/m3 across the same period. These trends are also apparent when these equal divided groups are mapped onto the different three major build
77 types (fig 3-5c). Collectively, these data confirm my original observation that newer properties contain greater radon.
Figure 3-5 – Residential radon concentrations divided by year of home construction. (a)Radon concentration charted on a ln axis according to year of construction quantile division (n=4996) (yellow; 0-99 Bq/m3 - 48%, orange; 100-199 Bq/m3 - 38%, red; ≥200 Bq/m3 16%. (b) Log transformed radon concentration plotted against year of construction with the linear regression line plotted (blue) (F (1, 7791 = 235.2, p < .0001). Radon Concentration increases with decreasing home age. (c) Radon concentration vs Build type, with subdivision based on quantile division of the year of construction division. The upward trend in radon concentration in newer homes is visible across the major build types.
78
3.3.4 Household Radon Concentration as a Function of Square Footage and Ceiling Height
In my initial survey of radon in Southern Alberta, I speculated that the surface area and net height of properties may be important contributors to higher radon in newer homes, which are demonstrably larger and broadly understood to be taller. To interrogate the effects of surface area and height of home, participants were asked to report their properties square footage (for basement and main floor), ceiling heights (by floor) and overall number of stories. I found statistically significant differences in the radon concentrations when homes were separated out by reported square footage brackets (p<0.005) (fig 3-6 a) In a subset of homes, a smaller basement surface area was reported by the homeowner compared to the reported ground floor surface area. In these homes, the analysis based on basement surface area still shows group differences in means, but with a lower level significance (p<0.05) (fig 3-6 b). In contrast, I saw no significant differences in indoor air radon levels as they relate to the number of stories / levels reported for properties (fig 3-6 c). However, an ANOVA analysis revealed significant differences between groups based on the reported ceiling height of the basement (p<0.0005), main floor (p <0.0005) and upper levels (p <0.005), respectively (fig 3-7 a-c). A multiple regression analysis predicted a higher radon concentration in properties with taller basement and main level ceilings (F (3, 4968 = 51.59, p < .0005)), with the coefficients indicating lower radon levels in upper floors with higher ceilings. Regression analysis was used to investigate the hypothesis that square footage mediates the effect of year of construction on radon concentration. Results indicated that square footage was a significant predictor of year of construction, b = 0.404, SE = 0.181 p < 0.05 and that Year of Construction wasn’t a significant predictor of square footage, b = 0.819, SE = 0.299, p =0.9. These results do not support the mediational hypothesis.
79 Figure 3-6 – Residential radon testing results divided by major dimensions of the building. Mean radon level (+/- 95% CI) based on reported home dimension metrics, omitting groups of less than 30 tests (ANOVA p-values). Pairwise post hoc testing details in Table 9. (a)Square footage of main floor (p<0.0005), (b) Square footage of Basement (p<0.05), (c) Number of Stories (p<0.05)
80 Figure 3-7 – Residential radon testing results divided by Ceiling Heights of respective floors Mean radon level (+/- 95% CI) based on reported ceiling height on different floors, omitting groups of less than 30 tests (ANOVA p-values). Pairwise post hoc testing details in Table 9. (a)Ceiling Height – Basement (p<0.0005), (b) Ceiling Height – Main (p<0.0005), (c)Ceiling Height – Upper (p<0.05).
81 3.3.5 The impact of occupant behaviour, other home metrics and the overall model
3.3.5.1 Radon as a function of structural attributes of the basement level and foundation
As radon primarily enters a property via its lowest level, I felt it was important to assess whether distinct types of foundation, basement floor (slab) or basement walls influenced
Figure 3-8– Residential radon testing results by attributes of basement (i) Mean radon level (+/- 95% CI) based on reported metrics describing the basement, omitting groups of less than 30 tests (ANOVA p-values). (a) Presence of walk out basement (n.s.), (b)Presence of Plumbing or crawlspace in the basement slab (n.s.)
82 Figure 3-9 – Residential radon testing results by attributes of basement (ii) Mean radon level (+/- 95% CI) based on reported metrics describing the basement, omitting groups of less than 30 tests (ANOVA p-values). (a)Foundation Type (n.s.), (b)Wall of Basement (n.s.), (b)Slab Type (n.s.)
83 the average radon within a given structure. Homeowners reported their foundation type as one of four classes typical of the region: crawl space, basement, slab-on-grade or bi- level. The slab type (concrete or earth), the presence of a walkout basement and whether the basement walls were wood, concrete or cinder block was also recorded. Neither the type of foundation, nor the type of slab or wall types in the basement correlate with increased or decreased radon exposure in our test group (fig 3-8 a-c) The presence of crawl spaces or a walk out basement also lacked any clear significance in radon levels (fig 3-9 a-b). In summary, none of these environmental design features impacted the ability of the property to draw up or retain radon.
3.3.5.2 Occupant behaviour influencing home air dynamics
Thermal stacking (hot air rising) has also been suggested to be a significant factor in drawing radon from lower levels to upper stories of a property, as it can generate greater negative pressures which can draw soil gasses out of foundations. In the questionnaire, I asked participants to indicate their typical thermostat settings during difference times of the day, conforming to the typical temperature range (<16oC to >23oC). Four different periods were considered: (1) daytime with the home unoccupied, (2) daytime with the home occupied, (3) evenings and (4) nighttime when occupants were asleep. Based on analysis of variance there was no significant difference in radon concentrations based on the temperature setting of the thermostat at any time of day (fig 3-10). The air dynamics of a property may also be influenced when occupants open windows. In some cases, this simply may contribute to air dilution effects as incoming outside air removes radon. However, depending on wind speed and direction, and whether one or more windows are open, air may also be draw out of a property in a dramatic and directional manner – much like opening a car window at high speeds. As this may increase or decrease the negative pressure of a property at lower levels (and thus radon entry), I assessed whether regular window opening behavior influenced radon levels. The questionnaire specifically asked whether windows were opened on respective levels of the home (basement, main or upper) and typically during four categories: (1) frequently – always (all times of year), (2) frequently – warm (only when it is hot), (3) sometimes and (4) rarely. Based on responses, and somewhat expected, most occupants
84 indicated basement windows were opened only rarely. By contrast, main floor windows were most commonly opened sometimes or when it as warm. Upper floor windows were reported to be opened the most, with two thirds of responses indicating they were opened sometimes-frequently. Interestingly, an ANOVA analysis showed a difference between the reported frequency groups on all levels of the home, indicating that radon concentration may indeed vary with window opening behaviour (fig 3-11). Significant differences in indoor radon were noted in homes where the occupant indicated regular opening of upper floor windows, versus those that did not have windows open or opened lower floor windows.
85 Figure 3-10 – Residential radon testing results across different thermostat setting behaviours Mean radon level (+/- 95% CI) based on reported thermostat settings at different phases of the day, omitting groups of less than 30 tests (ANOVA n.s. @ p>0.05) (a) Thermostat setting - during day while away (n.s.) (b) Thermostat setting - during day while at home (n.s.) (c) Thermostat setting - during evening while at home (n.s.) (d) Thermostat setting - during night while asleep (n.s.). 86 Figure 3-11 – Residential radon testing results across different frequency of window opening Mean radon level (+/- 95% CI) based on reported window opening behaviors on different floors, omitting groups of less than 30 tests (ANOVA p-values). Pairwise post hoc testing details in Table 9. (a) Windows open frequency - basement (p<0.05) (b) Windows open frequency - main (p<0.05) (c) Windows open frequency - upper (p<0.005)
87 3.3.5.3 Overall and Subset regression Models
It is noteworthy that in all groups there is still considerable variation and no single predictor had an r-squared value over 0.03, despite the size of the dataset. Even taking a regression approach, a model considering all available predictors only reaches an r- squared of 0.14, explaining at most 14% of the variation in the data. Such a model is also almost certainly over-fitted, and it would not be advisable to attempt to use it to make predictions as to whether a given property with a specific set of metrics will or will not have radon of a certain range. However, my efforts here are to elucidate some of the underlying trends so that this study might inform future investigations, as well as policy and design considerations to help prevent radon exposure and the associated negative health implications, including lung cancers. Pursuant to this, I ran a stepwise subset
Figure 3-12 – Subset selection in overall regression analysis (a-d) selection criteria and variance statistics of subset regression analysis (F (4, 1740 = 36.13, p < 2.2e- 16). Forward and backward subset searches were each carried out using all appropriate predictors (p=50), indexes of searches are cropped to 12 (out of 50) predictor model (p=4 model highlighted by red dot): (a-b) Bayesian Information Criterion by indexed predictors for forward (a) and backward (b) model selection searches, minimum in each case was at 4 predictors. (c-d) R squared statistic by indexed predictors for forward (c) and backward (d) model selection searches.
88 analysis on the data, using both a forward and a backward stepwise search method with all predictors, using the subset of all tests that were fully reporting on the survey. I used a minimised Bayesian Information Criterion (BIC) for model selection (fig 3-12 a/b), asking (within the regression models’ capacity for variable isolation) what are the main predictors contributing to the power of the model, whilst also being aware that that does not imply a precise model. Both these searches returned a 4-predictor model with Year of Construction, Build Type (Semi-Detached -Townhouse), Room of Test (Bedroom) and Thermostat Setting (Evening) being selected in the regression (F4, 1740 = 36.13, p < .0005) (fig 3-12 c/d),.
3.3.6 Examining short term (5 Day) radon testing
Real time radon reading varies considerably (Appendix E) but currently, radon testing falls into two distinct categories; a short-term testing option that lasts on average 2-5 days206 and a long-term testing option that last a minimum of 90 days176. While both testing options have distinct benefits, long term testing has been discerned as the gold standard, with Health Canada, among others27, strongly endorsing it. Favourability of the long-term testing arises from their ability to have a larger sample size affording a more accurate prediction of one’s radon levels206. Specifically, this larger sample size results in more resistance to readings variability as levels tend to fluctuate over time due to metrological and behavioral factors206, (appendix E) giving a more representative reading. This overall temporal variation represents a sinusoidal pattern often marked by lower radon levels in summer and higher radon readings in winter206. Furthermore, favourability of the long-term testing option stems from the inconsistency of the short- term test options206–208 as they only represent a snapshot of radon, which consequentially make them extremely susceptible to environmentally induced variability. By and large these fluctuations have not been systematically analyzed in a controlled manner, nor have
89 quantified comparisons between short- and long-term test accuracy been made. To do this, I deployed 5-day alpha track radon test devices for the latter 5 days of a side-by-side 90+ day alpha track test in homes all across Alberta and compared the readings.
Figure 3-13 – Long- vs Short- term testing, separated by testing location in home (a,b) 5 day testers compared to 90 tests in the same home, separated by testing location, linear fitted trendlines with (y intercept =0)(r-square statistic quoted for each line)
90 Figure 3-14 – Long- vs Short- term testing, separated by short term test date (a-c) 5-day testers compared to 90 tests in the same home, separated by time period, linear fitted trendlines with (y intercept =0), r-square statistic quoted for each line.
Variation in test precision was seen across distinct seasonal testing periods. For March 2018, during which winter conditions were prevalent in Alberta (fig 3- 13a), 5 vs 90+ day paired tests displayed an r2 of 0.85; by contrast, paired tests where the 5-day
91 short term test period was concluded after mid-April of 2018 (warm weather conditions) only had an r2 of 0.67 (fig 3- 13). To assess this further, I provided a second 5-day alpha track test to 75 homes who performed 5-day radon tests in March of 2018 (and which had strong agreement between that test and their 90+ day reading). These additional tests were deployed in the same location within the same home for 5 days in August of 2018, and readings were compared to the matching 5-day radon test reading from March. There was essentially no agreement between readings, with paired 5-day winter versus summer tests displaying an r2 value of only 0.035 (fig 3- 13c). When separating out the datasets for homes testing radon on the main floor versus the basement, results showed a further reduction in coefficient of determination. Short versus long term tests on basement level had an r2 value of 0.83, whilst the same paired tests only had an r2 of 0.77 for main or upper floors (fig 3- 13c). Collectively, these data point towards ideal and detrimental test periods for short term tests (5). Given this information, the applicability of short-term radon tests may be questioned in terms of providing reliable data to calculate human dose exposure, and potentially for economically costly mitigation work for properties. The desire for short term use stems from temporal constraints (1), as long-term testing rather obviously will take a while (4) and, in certain instances such as a home inspection during a real estate transaction, time is a limiting factor (1 week or less) and would require a faster testing process (1, 4). Since short term tests represent a solution to radon testing in short term windows, the participants within my study were surveyed to see assess their opinions surrounding radon testing and short-term testing within the specific context of a real estate transaction. A total of 3500+ individuals were surveyed. Of 3500+ individuals surveyed, it became apparent that determining a future properties radon levels constituted a priority only for some, with 36% of people reporting that they were very likely to ask for a radon reading before buying a home, and 31% being somewhat likely to do the same. Few people, however, were very likely to accept a short-term test during a real estate transaction (18%), and even fewer trusted short-term tests (6.14%). Additionally, many said they would re-test regardless of the short-term test results; with 46% re-testing if the short-term test results were low and 82% re-testing if the short-term test results were high. An important caveat to these observations is that this cohort is already
92 educated with regards to radon, having opted into this study. Although I should stress that this population had no access to the data presented in figure 10, they are likely to have gained knowledge from multiple sources (such as Health Canada) about the benefits of long-term testing. It is worth saying that a “radon aware” population appears to agree with testing modalities and behaviours that conform to best practice based on the data I have gathered and described here.
93
Table 3-1 - Radon Levels by Location of Testing
Geometric Arithmetic Standard n Upper Lower Mean Mean Deviation CI CI Semi-Detached 58 84 90 140 99 69 Townhouse Detached 90 117 114 957 124 110 Not Specified Detached 101 119 75 397 126 112 Split-level Semi-Detached 92 123 128 196 141 105 Duplex Detached 102 131 200 1789 140 122 2 story Semi-Detached 107 138 115 117 159 117 Side by Side Detached 108 143 141 174 164 122 3 story Detached 117 145 122 1249 152 138 Bungalow Basement 108 137 165 3556 142 132 Ground 85 111 109 906 118 104 Bi-Level 93 109 73 156 120 98 First 84 113 118 416 124 102 Second 74 94 69 34 117 71 Bedroom 106 136 114 675 145 128 Living room 94 122 113 1416 128 117 Non-living area 109 134 109 376 145 123 Other living area 101 131 181 2583 138 124
94 Table 3-2 - Radon Levels by Geographic divisions based on ED and major population centers
Geometric Arithmetic Standard n Upper Lower Mean Mean Deviation CI CI NW Calgary 93 127 158 2549 133 121 SW Calgary 100 128 110 2223 133 123 E. Calgary 100 121 86 1594 125 117 S. Alberta 113 154 244 1292 167 141 (excluding major cities) Edmonton 104 129 95 909 135 123 N. & C. Alberta 101 123 87 940 129 117 (excluding major cities) Saskatoon 111 132 84 538 139 125 Regina 203 296 303 491 323 269 Saskatchewan 158 227 264 845 245 209 (excluding major cities)
Table 3-3 - Radon Levels by Year of Construction
Arithmetic Geometric Range 99% n Mean Mean Min-Max CI 1800-1974 117 91 2007 15-2596 (111,123) 1975-1993 116 89 2046 15-2116 (110,122) 1994-2004 129 102 1906 15-1575 (123,135) 2005-2018 165 122 1834 15-7199 (152,178)
95 Table 3-4 - Radon levels by Home Dimensions
Radon Concentration (Bq/m3) By Square Footage ( n= 5,096) Geometric Arithmetic Standard Upper Lower n Mean Mean Deviation CI CI 500-999 sq. ft. 87 113 97 694 120 106 1000-1499 sq. ft. 100 127 112 2094 132 122 1500-1999 sq. ft. 104 131 120 1230 138 124 2000-2999 sq. ft. 109 143 262 889 160 126 3000-3999 sq. ft. 118 159 154 143 184 134 4000 sq. ft. or more 94 132 113 46 165 99
Radon Concentration (Bq/m3) By Square Footage of Basement ( n= 5,111) Geometric Arithmetic Standard n Upper Lower
Mean Mean Deviation CI CI 0-499 sq. ft. 71 104 115 96 127 81 500-999 sq. ft. 94 121 104 1001 127 115 1000-1499 sq. ft. 101 128 112 2082 133 123 1500-1999 sq. ft. 103 130 119 1050 137 123 2000-2999 sq. ft. 109 146 287 725 167 125 3000-3999 sq. ft. 114 155 160 115 184 126 4000 sq. ft. or more 93 131 113 42 165 97
96 Table 3-5 - Radon levels by Home Dimensions
Radon Concentration (Bq/m3) By Ceiling Height
Geometric Arithmetic Standard Upper Lower n Mean Mean Deviation CI CI Basement (n=4981) < 8 ft 96 121 102 881 128 114 8 ft 96 121 103 3046 125 117 9 ft 121 163 274 938 181 145 10 ft 106 132 92 116 149 115 Main (n=4972) < 8 ft 79 104.1702 92.84169 47 131 78 8 ft 93 117.585 100.8968 2566 121 114 9 ft 106 135.2455 117.9549 1267 142 129 10 ft 120 159.3547 163.6571 468 174 145 > 10 ft 107 143.9199 303.5608 624 168 120 Upper (n=3260) < 8 ft 90 107 63 67 122 92 8 ft 92 119 177 2323 126 112 9 ft 108 146 154 579 159 133 10 ft 100 122 88 155 136 108 > 10 ft 97 139 163 136 166 112
Radon Concentration (Bq/m3) By Stories
Geometric Arithmetic Standard Upper Lower n Mean Mean Deviation CI CI 1 story 109 142 123 133 163 121 2 story 114 141 119 1628 147 135 3 story 100 130 189 2109 138 122 4 story 91 114 90 235 126 102
97 Table 3-6 - Radon levels by Basement Attributes
Radon Concentration (Bq/m3) By Basement Attributes (n= 5,096)
Geometric Arithmetic Standard Upper Lower
Mean Mean Deviation n CI CI Foundation Type (n=4991) Bi-level 95 117 86 245 128 106 Slab on grade 82 118 134 82 147 89 Basement 102 131 155 4599 135 127 Crawl space 93 133 170 65 174 92 Slab Type (n=3979) Earth / Dirt / Rock 106 134 159 3938 139 129 Poured Concrete 105 154 211 41 219 89 Wall of basement (n=3963) Cinder Block 94 116 69 58 134 98 Poured Concrete 105 133 162 3760 138 128 Wood 113 146 112 145 164 128 Walk Out basement - Yes/No (n=5079) Yes 100 128.1961 124.7447 1030 136 121 No 101 129.5804 156.9984 4049 134 125 Crawlspace – Yes/No (n=3988) Yes 104 133.2418 162.2226 3594 139 128 No 103 127.9594 113.4844 394 139 117
98 Table 3-7 - Radon levels by Thermostat Settings Geometric Arithmetic Standard n Upper Lower Mean Mean Deviation CI CI 16oC or less 106 144 172 70 184 104 17oC 102 128 98 285 139 117 18oC 101 129 107 601 138 120 19oC 101 133 212 1614 143 123 20oC 101 129 120 1535 135 123 21oC 98 124 99 780 131 117 22oC 93 115 90 156 129 101 23oC or more 106 144 172 70 184 104 16oC or less 103 131 101 646 139 123 17oC 102 130 120 674 139 121 18oC 98 121 103 1070 127 115 19oC 103 131 128 768 140 122 20oC 102 140 275 809 159 121 21oC 101 127 111 572 136 118 22oC 95 122 94 314 132 112 23oC or more 91 113 105 78 136 90 16oC or less 93 128 116 32 168 88 17oC 99 135 166 73 173 97 18oC 106 136 118 265 150 122 19oC 101 128 99 535 136 120 20oC 101 128 112 1578 134 122 21oC 103 135 219 1557 146 124 22oC 96 122 102 822 129 115 23oC or more 93 112 84 183 124 100 16oC or less 98 123 94 764 130 116 17oC 101 129 118 863 137 121 18oC 99 133 236 1191 146 120 19oC 106 136 134 861 145 127 20oC 100 127 107 628 135 119 21oC 105 132 108 432 142 122 22oC 91 112 82 241 122 102 23oC or more 96 122 110 79 146 98
99
Table 3-8 - Radon Concentration by Frequency of Window Opening
Geometric Arithmetic Standard n Upper Lower Mean Mean Deviation CI CI Windows Open – Lower Floors (n=4443) Rarely 100 128 158 3772 133 123 Sometimes 109 145 152 527 158 132 Frequently (warm) 76 102 116 85 127 77 Frequently (always) 94 124 106 59 151 97 Windows Open – Main Floors (n=4828) Rarely 104 137 230 1229 150 124 Sometimes 104 131 115 2601 135 127 Frequently (warm) 84 114 125 657 124 104 Frequently (always) 96 117 84 341 126 108 Windows Open – Upper Floors (n=3581) Rarely 102 137 242 1112 151 123 Sometimes 100 127 109 1650 132 122 Frequently (warm) 82 109 122 410 121 97 Frequently (always) 78 101 95 409 110 92
100 Table 3-9 – Post hoc test (pairwise reporting) for radon level by home metric ANOVAs
Radon Concentration (Bq/m3) Differenc P< 95% 95% e from P< HOLM- lower upper mean B Square Footage 8.31290 51.8276 0.00116 0.01741 2000-2999 sq. ft. v 500-999 sq. ft. 30.07027 1 4 1 9 6.22509 85.1186 0.01246 0.17449 3000-3999 sq. ft. v 500-999 sq. ft. 45.67187 8 5 4 6 Square Footage Basement 46.4486 0.01312 2000-2999 sq. ft. v 500-999 sq. ft. 24.78806 3.12745 0.27559 7 3 Ceiling Height - Basement 23.1084 59.4274 9ft vs <8ft 41.26794 3.1E-08 1.55E-07 7 1 27.0506 55.9574 9ft vs 8ft 41.50406 0 0 8 5 Ceiling Height - Main 3.46332 31.8576 0.00623 0.04986 9ft vs 8ft 17.6605 7 8 3 2 20.9868 62.5526 10ft vs 8ft 41.76974 4.34E-07 4.34E-06 6 3 7.87949 44.7903 0.00094 0.00853 >10ft vs 8ft 26.33491 3 4 8 5 1.74326 46.4752 0.02716 0.19015 10ft vs 9ft 24.10924 8 1 5 4 Ceiling Height - Upper 5.36517 47.7804 0.00572 0.05723 8ft vs 9ft 26.57281 5 5 4 7
Geographic Area 9.66009 44.1954 0.00093 Southern AB. v. NW Calgary 26.92779 4.67E-05 7 8 3 144.682 194.521 Regina v. NW Calgary 169.6018 0 0 3 3 79.8200 119.962 SK v. NW Calgary 99.89117 0 0 8 3 8.42930 43.8062 0.00016 0.00307 Southern AB. v. SW Calgary 26.11776 2 2 2 9 143.578 194.004 Regina v. SW Calgary 168.7917 0 0 8 6 78.6469 119.515 SK v. SW Calgary 99.08114 0 0 3 4
101 14.1758 Southern AB. v. East Calgary 33.10366 52.0315 2.1E-06 4.41E-05 2 149.680 Regina v. East Calgary 175.7776 201.875 0 0 3 84.5510 127.583 SK v. East Calgary 106.067 0 0 4 1 0.01322 0.22476 Edmonton v. Southern AB. -24.7842 -46.6732 -2.89528 2 9 Northern and Central AB. v. Southern 0.00048 0.00869 -30.3409 -52.0169 -8.66484 AB. 3 1 Regina v. Southern AB. 142.674 115.868 169.48 0 0 50.5931 95.3336 SK v. Southern AB. 72.96338 0 0 3 4 139.139 195.776 Regina v. Edmonton 167.4582 0 0 8 7 73.5856 121.909 SK v. Edmonton 97.74762 0 0 2 6 144.860 201.169 Regina v. Northern and Central AB. 173.0149 0 0 6 1 79.3349 127.273 SK v. Northern and Central AB. 103.3043 0 0 8 6 195.671 Regina v. Saskatoon 164.1137 132.556 0 0 3 122.291 SK v. Saskatoon 94.40307 66.5149 0 0 2 SK v. Regina -69.7106 -98.4027 -41.0185 0 0
Build Type 6.62413 86.8465 0.25609 D - 2 Story-SD - Townhouse 46.73534 0.00985 9 5 1 6.72786 110.510 0.01433 0.35839 D - 3 Story-SD - Townhouse 58.61921 3 6 6 4 19.8549 101.326 0.00017 0.00499 D - Bungalow-SD - Townhouse 60.59061 6 3 9 9 47.3369 0.00050 0.01373 D - Bungalow-5 27.70181 8.06665 6 9 8 Floor Type 0.00027 main v basement -26.19 -41.4254 -10.9546 2.75E-05 5 0.01698 0.15290 first v basement -24.0513 -45.2647 -2.83791 9 1
Windows Open Frequency - Main 0.01244 0.07466 Frequently (warm)-Rarely -22.4333 -41.3551 -3.51144 4 2 Windows Open Frequency - Upper
102 0.01717 0.08588 Frequently (warm)-Rarely -27.5903 -51.6775 -3.50307 7 5 0.00091 0.00550 Frequently (always)-Rarely -35.453 -59.5617 -11.3443 8 6 0.02457 0.09831 Frequently (always)-Sometimes -25.306 -48.3336 -2.2784 8 3
3.4 Discussion
3.4.1 Summary / Interpretation
This is a snapshot of an ongoing study, with close to twelve thousand long-term radon concentration tests, most with up to 50 linked home metric and behavioural data points, as well as the numerous short-term tests and publicly available statistics that have been used to map and analyse the data an attempt completely characterise the data and link out its myriad implications if a Sisyphean enterprise.
3.4.1.1 Build Types
The data covers a wide area of analysis but successfully samples across the population, with more than 30 tests in all but three of the EDs of Alberta and Saskatchewan. Sampling was higher in urban areas than rural and was higher in AB than in SK; however, this does match the population dynamics and the sampling exceeds any previous efforts. Notably, the Health Canada cross-Canada radon survey, for the same area, only monitored 2,411 tests compared to the 11,276 I describe here. In our results, 18% of tests recorded radon levels at or over 200 Bq/m3, contrasting starkly to the overall national estimate of 7% from the Health Canada study. The reasons for this discrepancy are not obvious considering all our tests carried similar methodologies (test types, periods, and lengths) and, in fact, were based on the HC methods. One caveat to the original work from Health Canada is that it was unclear whether all participants followed the instructions and removed the radon test device from the sealed mylar plastic bad in which it arrives. A close examination of the raw dataset from that study (obtained through the Freedom of Information process) reveals multiple readings at 0 Bq/m3 which is somewhat unusual and perhaps indicative of failed deployment. From our results, AB has 14% of properties with radon above the national maximum exposure threshold, while
103 35% of SK homes were above it. While it seems clear that the higher SK numbers push our overall estimates higher for the Prairies, these numbers are population weighted appropriately, and so suggest “high” radon is bigger issue than previously reported (14% above 200 Bq/m3 in AB here, vs 7% in the Health Canada study). Potential specific influences that may explain the differences between my work and that of Health Canada are explored in the following sections, however the main difference in the studies is the period of testing (discussed in section 3.5.1.2) and the sampling methods. The Health Canada study recruited over the telephone via a contracted market research firm, whilst our study recruited through participant-directed online signup following organically generated public relations, media exposure and social media advertising. Health Canada issue free radon kits, whilst the participants in my study operated as citizen scientists and paid for their devices. While these distinct practices may have skewed the participant populations (in either study) towards certain demographics, it would be hard to see how these could be so dramatically different as to influence the dataset to a significant degree, especially considering there are no marked selection criteria stated in the HC study that differ from our own (e.g. excluding high rise and rental properties from analysis). Further, human demographics are less relevant here compared to the property demographics, which (as described in the previous chapter) broadly conform to the means for the survey region. If demographic data or geographic data from the other studies were available, I could have attempted to compare our data, however such data is not available to us nor did we ask to collect human demographic data in this study. It would be possible to approximate some demographics from the geospatial information in our dataset; however, because of the scale of such analysis and the lack of a useful comparison, that remains outside the scope of my work. Results of radon tests placed on different floors and/or rooms of testing support the widely held notion that radon levels are highest at the lowest levels of a property. At the same time, my work also indicates that radon levels for any location above ground floors are broadly the same within a given property. This is likely due to air-mixing effects connected with internal air ducts flowing mostly in a unidirectional manner from basement to upper floors, and air-dilution via exterior windows and doors which will be predominantly on upper floors.
104
3.4.1.2 Differences in Radon Concentration Levels by Year of Construction
The year of home construction appears to have a significant predictive capacity on the radon concentration, although there is considerable variation on the data; thus, the prediction is accurate but imprecise. This may have implications in cities with rapidly expanding new housing stocks, drawing the question is radon likely to become an ever increase contributor to lung cancer. This may also partly explain some differences between the HC study and my work. As a reminder, the Health Canada study estimated 6.6% of AB homes and 16.3% of SK homes in their 2007-2009 survey had radon concentrations of ≥200 Bq/m3, whilst my work (based on a much larger sample size) spanning 2011-2018 indicates this is 14% and 35%, respectively – basically double what was previously understood in each case. In our dataset of homes accurately reporting year of construction (n = 4,996), those homes built on or before 2009 display slightly lower risk of exceeding the maximum acceptable radon exposure threshold (12%) than those built since 2009 (26%). It is not clear from this whether the differences seen between studies can be explained by the testing periods or if it is a home age effect. There may be other environmental design considerations that are the reason for these trends, and the fact that different types of buildings have differing radon concentrations supports this concept. Indeed, pairwise testing in build type data revels that this effect may be a result of the lower radon levels seen in townhouses and semi-detached build to fully detached builds. I previously hypothesized that several more specific environmental design metrics may have an influence on this effect, including increased surface area (square footage) in newer homes, increased ceiling heights in newer homes, tighter air seal in newer homes and altered concrete shrink193. In the subsequent phase of the study, I took the opportunity to survey several of these metrics, specifically surface area and ceiling height (3.4.4); however, I found no evidence supporting the hypothesis that these dimensional metrics were mediating the home age effect. The hypothesis cannot be entirely rejected, however, as the considerable variation in the radon concentration may mask the effects from statistical interrogation even in a large dataset. Notably, my dataset includes a variety of home build types, and the home age effect is consistent across all types of construction ranging from single family detached bungalows to multi-storey townhouses and others
105 typical of the Canadian Prairies (fig 3-5c).
3.4.1.3 Geospatial Differences in Residential Radon Levels
I found that, even within the most precise geographic divisions of Prairie regions, there is considerable variability in radon test results, and no areas tested consistently low for indoor air household radon. This supports the necessity for individual households to continue testing for residents to know their own exposures and risk. Although human radon exposure is issue with a geological origin, it is influenced by other factors including environmental design and property engineering, occupant behaviour and seasonal weather patterns. Having said this, it does seem that is substantially higher radon in the City of Regina region. On a more localized level of comparison, 25.5% of homes (n=90) in the Regina Qu'Appelle Regional Health Authority region reported above 200 Bq/m3 in Health Canada’s 2012 Cross-Canada survey analysis. In my analysis of this region, 50% of homes (n = 491) were ≥200 Bq/m3. I should note, however, that the specific electoral divisions I used are not identical to the health authority region districting used by Health Canada, although there is broad overlap176. Clearly, my work has superiority in terms of sample size (being five times larger), but it is unclear whether that alone is enough to explain the 2-fold differences consistently being observed between these studies in terms of radon dosimetry observations.
3.4.1.4 Metrics of Environmental Design Dimensions
Based on the analysis of square footage, it appears that the increase in reported property surface area does indeed influence indoor air radon levels. In pairwise analysis, larger square footage groups had higher radon levels (e.g. 2000-2999 sq. ft. v 500-999 sq. ft. difference = 30 Bq/m3 (6,85), p<0.05) (table 3-4). This fits with my previous hypothesis that radon entry from the soil underneath a home is exacerbated when the contact area between the soil and property is increased. Whilst the overall number of levels in a property did not impact indoor radon, the increasing height of the basement and main floors did correlate with higher radon (e.g. main floor 10ft vs 8ft, difference = 41 Bq/m3 (21,63), p<0.0001) (table 3-5). This is in line with hypotheses that taller ceilings enable
106 more powerful thermal stacking effects (i.e. hot air rising further) to generate greater negative pressures on the lowest levels, which draw greater radon into a property. None of the investigated attributes of the basement (other than square footage, discussed above) appeared to have any large effect on radon concentrations (table 3-7). For these metrics, one caveat I acknowledge is that it is possible that that some homeowners are unable to accurately identify each metric (especially if their basement was developed), a common limitation of self-reported data. Saying that, the presence or absence of a walkout basement is very unlikely be inaccurately reported and the group means did not significantly differ. So, while this does not prove the negative, it does reject the hypothesis that this markedly different basement construction has a significant influence on indoor radon concentrations.
3.4.1.5 Occupant Behaviours and its impact on Radon Levels
Whilst the fundamental environmental design metrics of a property can strongly influence indoor air radon concentrations, there is also the potential for human behaviour to have a significant influence. In theory, internal air temperatures should influence thermal stacking within a property, with hotter indoor temperatures potentially creating a larger pressure differential at lower levels as the hotter air rises and escapes the structure (typically through the roof). By this same logic, opening windows on lower, ground or upper floors may also impact pressure differentials, with windows opened on upper floors creating direct air escape routes that theoretically would draw more air from lower levels. Opening basement windows, although likely a rare behaviour for most homeowners, could de-pressurize upper floors. To address this, I surveyed participants and asked them to indicate typical thermostat temperature settings during difference times of the day, and the frequency of window opening (and on what floor). In summary, the results indicate that radon concentration may vary with window opening behaviour, although not the actual temperature the internal air is balanced at. This makes some sense, as pressure differentials generate by active air suction through windows is likely to be stronger versus any increase in passive air diffusion generated by thermal stacking within the temperature range typical of a Canadian home (16-23oC). It is possible that taking external temperatures into consideration as well as thermostat settings may be necessary to see
107 effects of these setting on indoor radon concentrations, as it is the temperature differential not the absolute temp that drives the pressure differentials, such analysis would require meteorological data as well as real time radon concentration reading which were not collected in sufficient number for analysis in this work.
3.4.1.6 Overall model
Using all metrics collected to date to try to predict radon concentrations does not offer many insights, other than to confirm that there remains a great deal of variation in the radon concentration that is not accounted for by these predictors. The subset selection models offer some leads as to which metric most increase the predictive power of the model, although this does not increase the accuracy it is way of ‘training’ the model and omitting less relevant metrics. Both the forward and backward selection models landed at the same predictors. The coefficients for this 4-predictor model [year of construction (7.8e-3), Semi-detached Townhouse (-6.4e-1), Testing Location – Bedroom (8.3e-2), Thermostat Settings – Evening (-2.6e-2) are relevant to a log n transformed model and so are not immediately quantifiably interpretable; however, there are some qualitative insights. Firstly, the selection of year of construction and its positive coefficient are not surprising since the newer homes had higher mean radon in our other tests. A dummied predictor, Semi-detached Townhouse is not surprising to see here either, and its negative sign supports the lower mean I saw in this group in the Built Type ANOVA. The third predictor, another dummied one, Testing Location – Bedroom did not show up in as significant in the metric ANOVAs, nor did the fourth predictor Thermostat Settings – Evening. It is not clear why testing a bedroom would add significantly to the power of the model; however, the coefficient is positive implying a positive correlation with radon levels. The thermostat setting predictor selected had a negative coefficient implying that higher temperature settings in the evenings negatively correlate with radon concentration. I would urge caution in any over-interpretation of these models, and support further investigation into environmental design metrics, rather than claiming causative proof of their relationship to indoor radon levels at this stage.
108 Chapter Four: High-throughput, benchtop alpha particle irradiation system for the investigation of high linear energy transfer radiation in biological systems
4.1 Introduction
As outlined in detail within the introduction to this thesis, LET is a measure of the energy deposition rate of IR. High LET IR is qualitatively different from low LET IR and has markedly different effects on living systems particularly CNS. Alpha particles are a form of high LET radiation that result from radioactive decay of unstable elements, such as 222Rn, as well as being a predominate part of cosmic radiation. Alpha particle radiation are particularly damaging and mutagenic, as they cause denser clusters of DNA damage in more restricted space 197. These clustered lesions are difficult to repair with fidelity and can lead to durable mutations in cells and which can be transmitted to subsequent cell progeny if the cell is dividing. These mutations can lead to cancer, and in fact, radon gas inhalation and the subsequent alpha particle-mediated DNA damage is the second leading cause of lung cancer (estimated 16% of cases) 27,159,193. Photon-based x- and γ- rays are qualitatively different from particle-based radiation, such as alpha particles, neutrons and protons. The ICRP Report 103 describes the biological weighting of alpha particles as 20, compared to 1 for photon-based X and γ IR 26. This relative biological effectiveness is an important consideration in investigating the cellular effects of IR; however, very little is known about the repair of high LET- mediated DNA damage, and its effects on cellular health. There is evidence to suggest that the repair of high LET-induced DNA damage is slower than low LET and relies more on homology-directed repair processes for resolution157,197,209. There is also literature to suggest that high LET radiation causes large-scale chromatin changes and inaccurate repair causes large-scale chromosomal rearrangement 197,210. Human exposure to high LET IR is derived from high mass and energy cosmic radiation, and medical procedures such as proton radiation therapy that incorporate increasing amounts of higher LET IR. Whilst the relative biological effectiveness of alpha particles is much higher than that of protons, there is evidence that suggests the repair, cellular impact and signatures of proton-mediated DNA damage is very similar to alpha particles 209,211,212. The efficacy of
109 protons in anti-cancer therapy is quickly making it a mainstay of cancer therapeutics, which further highlights the need to better understand how this therapy impacts the health and viability of not only cancer cells, but healthy tissue that also receives IR. There is a distinct lack of research on the cellular effects of high LET IR. In the 21st century, published studies (searchable on PubMed) with “high LET radiation” as targeting search term represents only 1.06% of all literature around radiation and human cells. This lack of research productivity is almost certainly due to the practical, economic and technological difficulties associated with studying high LET IR in a laboratory setting. Traditionally, an investigation of this type of radiation has required either hugely expensive particle accelerators to generate particles of enough energy, or the production and storage of dangerous and impractical radioactive gasses. Here, I have described and validated an easy, affordable and novel method to investigate high LET IR at the benchtop in any standard laboratory setting. My novel setup uses residential smoke- detector derived 241Am pellets attached to a custom, 3D-printed peg-and-scaffold system that brings the pellets in close proximity with cells in a 96 well plate. 241Am is a synthetic radioisotope that does not occur in nature and emits alpha particles of 5.486 MeV energy as well as a gamma emission of 59.540 keV with a half-life of 432 years (thus is also a ‘functionally stable’ radiation source). This alpha particle energy compares closely to those emitted during radon decay (5.489 MeV) (fig 1-3), making it ideal to use to model and discern the biological effects of IR emitted by decaying radon. This system represents a significant advance over a previously published method213 and any previous particle accelerator technology for multiple reasons: 1) It eliminates the need to culture human cells on Mylar plastic film. 2) It brings a medium-to-high throughput aspect to this method within a 96 well plate format and avoids the need to obtain highly limited restricted ‘beam-line’ time. 3) It only requires commercially available and inexpensive materials. 4) It is consistent in dose accuracy and precision and is adaptable to multiple assays and genetic model systems (beyond mammalian cell culture).
110 5) After irradiation, multiple parameters of the DNA damage response can be examined and quantified with our purpose built, TANGO based 3D nuclear analysis pipeline and other assays can be used to examine cell viability.
4.2 Results
4.2.1 Design and dose quantifications of the alpha particle irradiation system
To innovate on a previously described α particle IR methodology using a 241Am pin point source213, I set out to develop a methodology that was compatible with medium-to-high throughput laboratory plastic and glassware common to most biology labs worldwide. 96 well plates with microscopy-grade optical glass bottoms were selected based on the dimensions of the 241Am sources. Using computer aided design and 3D printing, I designed a custom bracketing-and-scaffold system for the irradiation of mammalian adherent cells growing on the surface of a 96 well plate. The bracketing system design brings the radioactive alloy to within 0.1 mm from the glass surface, ±0.01 mm due to variability in the construction of the steel housing for the 241Am sources themselves. To internally duplicate any condition, my design is based on a two-source bracket (fig 4-1 a,b) with one 241Am source adhered to end of each peg. This design is flexible and easily adaptable, and I also designed an 8-pronged bracket so that an entire column of the plate could be irradiated in synchrony (fig 4-1c ,4-8). These brackets can be removed and inserted from the overall scaffold system individually, and this allows for different doses and different exposures to be changed in parallel. Within the bracket and scaffold design, I then examined the dosimetry of the 241Am sources in the 96 well plate setup, to calibrate the system. To achieve this, I exposed the 241Am sources to EBT3 GAFchromic film in an orientation and manner strictly comparable to the 96 well plate set up. EBT3 GAFchromic film is a self- developing, high-resolution film that undergoes a color change proportional to the amount of absolute radiation received. As the active layer of the EBT3 is sandwiched between two 125 μm thick polyester layers, the top-most polyester layer was removed to expose the active layer to alpha particles without any hindrance (Mukherjee 2015) (fig 4-1 d). The film responded in a dose dependent manner to IR exposure and were compared to
111 pre-irradiated known standards, demonstrating a dose delivery rate of 0.4 ± 0.08 Gy per minute. This is a significantly lower dose rate as compared to those achievable with our GammaCell 137Cs source (2.8 Gy/min ± 0.4) or LINAC (100 Gy/min ±0.01) irradiation systems, which was desirable to experimentally mimic the more common modality of environmental high LET IR exposure; low dose with chronic exposure (appendix D). For an alpha particle of 5.46 MeV from the 241Am, the approximate path length in water is 65 μm, and the range in air is 6.33 cm. In this system, a cumulative dose of 0.8 Gy was delivered over a 2-minute period to cells exposed to air after media removal, the dose was reproducible by repeated measure (average percent difference 0.0% ± 3.6%). There was slight variability, 0.32 – 0.47 Gy/min (min-max) in dose delivery between sources, broadly comparable with the variation in other laboratory-grade IR sources (such as the GammaCell 137Cs source, which has a 14% dose variation from sample to sample). With the corrective power of internal duplicates (or even greater n) in evening out sample-to- sample variation, this technology allows for a sufficiently precise dose rate to enable experimentation.
112 Figure 4-1 – Design and Dosimetry of 241Am Irradiation setup (a) Schematic of the components and setup to irradiate cells for immunostaining. Cells are plated in a glass bottom 96wp, onto of which is paced a bracket unit that matched the well layout and ensures the correct spacing, 241Am sources are mounted on source holders which are then lowered into wells for irradiation of cells, this setup is designed to deliver equal doses to 2 wells at a time creating internal duplicates of doses. Dose is controlled by time of exposure. The bracket and source holding components can be modified as desired through computer aided design and 3D printing. (b) Images showing 241Am sources mounted on 2 prong source holders beside a 2x8 bracket and (c) 241Am sources mounted on 8 prong source holders assembled in a 6x8 bracket and a glass bottom 96 well plate. (c) A 3 channel RGB scan of EBT3 GafChromic film after 6min exposure to 8 different 241Am sources at equivalent distances that cells are exposed. 113
4.2.2 γH2AX induction, high-resolution image acquisition and data analysis pipeline
With the dosimetry of the 241Am sources successfully calibrated, I began to investigate the cellular effects of α particle IR on the biological responses in human cell culture systems. To do this, I first examined the DSB response by quantification of the surrogate DSB marker phosphorylated histone 2A variant H2AX (γH2AX). Having established a procedure to irradiate in a reproducible manner, I used quantitative image-based cytometry to analyse the nuclear signalling resulting from DNA damage after α particle
Figure 4-2 – Example Images of alpha particle irradiation (a) Confocal microscopy image of a 48br cell irradiated from the side with a point source, fixed at 1hr: ɣ- H2AX (green) and DAPI (blue). This irradiation demonstrates the linear track of damage caused by 241Am alpha particle emission. (b) Widefield microscopy image of a 48br cell irradiated for 2min alpha as per the top down 96wp protocol, fixed at 1hr: ɣ-H2AX (green) and DAPI (blue). This image shows the complex clustered ɣ-H2AX signal that cannot be easily counted as individual foci. (c) Widefield microscopy image of the entire surface of a well in a 96wp. The well contains 48br cells irradiated for 2min alpha as per the top down 96wp protocol, fixed at 1hr: Color represent Red-Blue LUT of the ɣ- H2AX channel. Low signal(blue) in the ɣ-H2AX is seen along the perimeter of the well, while higher signal (green) is seen in the center within the irradiation field, the edges of the filed are apparent.
114 IR. As outlined in detail within the introduction, γH2AX can propagate across megabase distances from a DSB and is a well-established marker of the DSB signalling response214. The induction of γH2AX is readily demonstrable in human cells after exposure to IR from an 241Am source. Previous reports have shown this using a custom manufactured 241Am pin-point source set at a perpendicular angle to the cells in order to generate a “track” of DSBs induced by alpha particles. I reproduced this angled exposure (fig 4-2a), however, a caveat with this single source method is the low throughput nature that restricts
115 (a)
Figure 4-3 – Automated image analysis, workflow and examples (a) Workflow of image analysis, multi-channel z-stack images of a field of cells acquired on a widefield microscope are imported into ImageJ/TANGO. Individual nuceli are isolated and the channel for subnuclear signal (here ɣ-H2AX) is analysed across the z-plane and structures are segmented out, before measurement. (b) Representative 48br cells irradiated for 2min alpha, fixed at 1hr: ɣ-H2AX (green) and DAPI (blue). Image acquired through the automatic acquisition of the widefield microscope at 40x objective, with 5 z-stacks. (c) The segmented 3D volume of ɣ-H2AX signal isolated from the cell shown in (b)
investigation to only an impractically small number of conditions/samples at a given time. While my top-down irradiation system lacks the visually impactful tracks that can be imaged with confocal microscopy and angled irradiation, deconvolution can produce
116 visual demonstration of the clustered DSBs thought to be induced by high-LET IR (fig 4- 2b). An added benefit if our setup against above the point source is that many cells can be simultaneously irradiated, even within a single well hundreds of cells can be irradiated with a similar (dose fig 4-2c). Following α particle irradiation, I used widefield immunofluorescence microscopy to analyse nuclear γH2AX signalling215,216. To do this, an automated imaging protocol was designed, and a 40X air objective was used to capture five z-stack images based on
Figure 4-4 – Image analysis Outputs 48Br cells exposed to 241Am radiation for the designated time and allowed to recover for 1hr. Cells were fixed, stained with an antibody to detect ɣ-H2AX, and ɣ-H2AX was automatically imaged and analyzed by individual cell, a panel of measurement outputs are displayed. Note: dose rate of 241Am sources was measured at 0.4 ± 0.08 Gy per minute. (a) Average number of ɣ-H2AX objects counted per cell nucleus. Based on automated segmentation(ɣ-H2AX). Peaks and then decreases, perhaps due to objects merging at higher doses. (b) Average intensity of ɣ-H2AX channel per cell nucleus doesn’t use segmentation of ɣ- H2AX channel, reports the average total nuclear intensity of the channel. (c) Average volume of ɣ-H2AX objects per cell nucleus, based on automated segmentation(ɣ-H2AX). Takes all objects found with a single nucleus and adds the volumes as opposed to (a) where they are simply counted. (d) Average volume of ɣ-H2AX objects per cell nucleus as percentage of the nuclear volume. Based on automated segmentation(ɣ-H2AX). Takes all objects found with a single nucleus and adds the volumes as in (c) and then corrects for the segmented nuclear volume (DAPI channel).
117 Nyquist sampling optimums, with approximately 500 nm stack shifts. The whole well (within the 96 well plate) could potentially be imaged using a tiling process; however, as the dose rate drops at the edges and there is an unirradiated “ring zone” of cells around the edge of the well, I preferentially imaged close to the centre of the well (dose fig 4-2c). Once the z-stack images were acquired, I used the TANGO plugin for ImageJ to process the images en masse and create a pipeline for the analysis of nuclear and sub-nuclear structures. TANGO is a free tool based on ImageJ and R functionalities, both of which are freely available, versatile, and are widely used in the scientific community. The use of TANGO allows for the modification of the methodology to the nuclear signal of choice (see (Ollion et al., 2013) for further instructions) (dose fig 4-2a-c). Here, I used the “Nuclear edge detector” and modified “Spot detector 3D” operations coupled with the “Measure geometrical, simple” and “signal quantification” modules to segment and measure the nucleus and model the γH2AX structures through the nucleus respectively (dose fig 4-3a-c). I used a TANGO pipeline to extract the (i) average number of segmented γH2AX objects per nuclei, (ii) the summed average intensity of each of γH2AX staining within segmented volumes per nucleus, (iii) the total volume of γH2AX segmented objects per nucleus and the (iv) total integrated density of γH2AX segmentation (dose fig 4-3a-c). Most of these measures peak at 2 hours post IR and decrease thereafter, representing the broadly understood repair kinetics of the high LET IR-induced DSBs. However, integrated density represents an aggregation of the other three measures, and thus I chose to report this as my principle value. TANGO can also be used to report nuclear measurements, such as DAPI integrated density (a measure correlated to total DNA content) and nuclear volume, and it can be informative to use these measures to ask questions about cell cycle staging or nuclear size variation215 (Appendix B).
4.2.3 α particle irradiation induced DSB repair kinetics
To further display how this irradiation method can be applied experimentally, I used it to comparatively analyze the γH2AX kinetics post irradiation in a high-LET and low-LET IR context. By varying the amount of time cells are exposed to the 241Am sources, a dose course of high LET radiation was developed (fig4-4d).
118 Based on these results, I picked the 2-minute exposure time, with an estimated 0.8 Gy absorbed IR dose, and ran a comparative time course with 0.8 Gy of x-rays (low-LET IR). Here, the γH2AX/DAPI integrated measure shows the DSB signalling peaks at 30 min for X-rays, substantially earlier compared to the 2-hour peak observed for high LET α IR (fig 4-5a)(Appendix C). Unsurprisingly, there is also a persistence in damage seen in high LET IR at the longest timepoint (24 hr post IR) compared to low LET irradiated cells. These dynamics match the expectations from previous reports using HZE particle irradiators217–219. This result validates my approach in the context of the established literature for high LET IR induced DSB repair and indicates that this method as a robust and accessible way to investigate the biology of high-LET IR exposure. These dynamics are also reproduced here in A549 lung adenocarcinoma cells, demonstrating the adaptability of the technique to other commonly investigated cell types (appendix C). To address the concern that the unavoidable 2 min of exposure of the cells to air may be contributing to the observed cellular effects via oxidative stress, a control time course with “sham” exposure (air exposure but no 241Am source exposure) was carried out. γH2AX did not change significantly from baseline levels during the 2 min period of air exposure. It is worth noting that, during this period, the cells are not “dry”, as a thin monolayer of aqueous media remains associated with the cell monolayers. Whilst clearly insufficient to block alpha particle bombardment of the cells underneath, this layer is enough to prevent dehydration and desiccation effects on cells (fig 4-5a-c). Saying this, my recommendation is that, in the future, this should be re-confirmed in any other cell types used in conjunction with this technology as different cell types may respond differently to air exposure / oxidative stress.
119 Relative γH2AX signal γH2AXRelative signal
Figure 4-5 – 48Br cells DSBR in an asynchronously population after alpha IR compared to gamma IR Alpha irradiated 48Br cells (red: 0.8 Gy), gamma irradiated cells (blue 0.8 Gy), and mock irradiated (grey: air exposed) and allowed to recover for the designated time periods. Cells were fixed, stained with an antibody to detect ɣ-H2AX, and ɣ-H2AX was automatically imaged and analyzed by individual cell. Error bars represent standard error of the mean. The plotted value considers the volume and intensity of the ɣ-H2AX and DAPI channels. (b) Representative 48br cells irradiated for 2min alpha, fixed at 1hr: ɣ- H2AX (green), 53BP1 (red), and DAPI (blue). (c) Representative 48br cells mock irradiated for 2min (air exposure), fixed at 1hr: ɣ-H2AX (green), 53BP1 (red), and DAPI (blue). 120 Figure 4-6 – Alternative Setup for Alamar and Comet assays Schematic representation of the components and setup to irradiate cells for plate-based assay (Alamar) or cell pelleted based assays (comet). Cells are plated at the center of the well and allowed to adhere before the well is filled with media, this ensures that the whole population of cells is within the desired radiation field. The glass bottom 96wp with center plated cells is then assembled with a bracket unit that matched the well layout and 241Am sources mounted on source holders. This setup is designed to deliver equal doses to 8 wells at a time creating internal duplicates of doses, this can be necessary for assays were cells must be collected as the number of cells required cannot be harvested from a single well. Dose is controlled by time of exposure, and if assay sensitivity limits call for higher doses, these doses can be delivered in fractions to avoid desiccation.
4.2.4 Quantification of high LET α particle irradiation induced DNA damage by alkaline comet assay, and effects on cell viability
4.2.4.1 Alkaline comet assay
To further investigate and quantify α particle IR-induced DNA damage, I used the alkaline comet assay as a direct physical measure of DNA damage induction – specifically DNA single and double strand breaks. In the context of low LET IR, the number and rate of γH2AX resolution serves as a robust surrogate readout for DSB number and repair214. However, I felt that it was important to examine the levels of DNA damage accrued by α particle IR in a more direct, physical manner, to ensure that the γH2AX assay was indeed applicable to this kind of high LET radiation. One method to directly visualize and quantify DNA damage is the comet assay, which, when run under alkaline conditions, reveals both single- and double-strand DNA
121 breaks and, when run under neutral pH conditions, is selective to DSBs only (fig 4-9a). The alkaline comet assay is a simple, physical readout of DNA damage, and relies on a series of detergents and alkaline salt environments to strip away cellular components down to the fragmented DNA, which creates a comet “tail” upon exposure to electrophoresis220,221, it is however not a very sensitive assay221. The comet assay has a linear relationship between IR dose and the comet tail moment (essentially the ratio between the DNA signal present in the comet “head” and the “tail”) and, as a single-cell analysis, reveals heterogeneity in DNA damage and repair within a cell population and across internal replicates and experimental repeats. The three main considerations when adapting our new high LET IR technique to traditional physical assays of DNA damage are the (i) dose rates, (ii) comet assay detection and resolution limits, (iii) dose distribution within a single well. The detection limits of the comet assay are about 50 strand breaks222, which corresponds to approximately 3-5 Gy of low LET γ-ray IR. Indeed, a neutral comet assay typically requires 30-50 Gy to observe DSB induction, and thus is substantially less sensitive when contrasted with IRIF enumeration which is sensitive (in a practical sense) down into the mGy range of dose exposure. However, alkaline comet assays require about 5-15 Gy to resolve DNA damage and are thus more sensitive222.
With the dose rate of the existing setup for the 241Am sources being 0.4 Gy/min, I utilized a fractionation method to avoid desiccation from extended media removal times >2 min. As only the center of each well receives the full 0.4 Gy/min, cells required precise plating in the center of the well to ensure an even spread of alpha particle exposure within the population of cells and precise dose distribution. To calibrate the comet assay, 48BR primary human fibroblasts were either left untreated, treated with DMSO or to 2.5 μM Olaparib (PARP inhibitor) before being irradiated with 15 Gy γ-ray IR; alkaline comet tails were examined up to 1 h post IR (fig 4-9a. The tail moment from 15 Gy γ-ray IR was nearly completely resolved by 1 h (fig 4-9). As shown in figure 4-9, the alkaline comet assay confirms the presence of DNA damage following eight 2-minute fractions of α particle IR, increasing the comet tail moment to 11.43 ± 0.66 μm (mean ± SEM) relative to the 1.88 ± 0.18 μm comet tail seen in the 0-fraction sham condition.
122 After 1 h, the tail moment had resolved nearly to baseline, although the kinetics of this were markedly slower compared to data obtained from γ-ray exposed cells. These data confirm the presence of α particle IR-induced DNA damage after exposure to 241Am, that is detectable by physical means, and resolves in a time-dependent manner. These data corroborate my earlier findings with γH2AX detection and quantification. Importantly, the comet assay has not historically been compatible with high LET radiation methodologies, and so my novel methodology opens a considerable range of experimental options to interrogate radiobiological problems with the comet assay, it also shows the capacity to use the setup for experiments where collecting a cell pellet is necessary. 4.2.4.2 Alamar blue cell viability assay
To quantify the effect of α particle IR on cell survival and viability, I performed a series of Alamar blue assays after fractionated α IR doses (fig 4-7). Alamar blue is a resazurin dye which, when reduced within healthy, metabolizing cells, fluoresces as a pink resorufin dye. This assay can be used to determine the concentration of viable (i.e. metabolically active) cells in a sample and is often used in conjunction with an analysis of growth kinetics of a cell line. I should note that the Alamar Blue assay lacks the exquisite sensitivity of the more complex clonogenic survival assay, where differences may be observed even following mGy exposures. However, what the Alamar blue assay may lack in dose sensitivity, it makes up in terms of speed – requiring only 1-2 days versus the typical 2 weeks required for a single clonogenic survival experiment. Figure 4-9 shows that fractionated doses of α particle IR result in dose-dependent reductions in cell viability, with a 40% loss of viability observed 32 hours after 16 α particle fractions, equating with approximately 12.8 Gy of α particle IR relative to sham treated controls. Strikingly, differences in viability are visible with as little as 1.6 Gy (two 2-minute fraction) (fig 4-7b). The fork system, setup and experiment were readily adaptable so that matched IRIF controls could be produced in a parallel well to those cells used for the Alamar assay. This is important, as it allowed me to quantitate γH2AX each time, in order to better confirm the IR dose and experimental conditions of the viability analysis (fig 4-7c). These doses are an order of magnitude higher than those typically used for a γH2AX IRIF time course, and so the γH2AX signal is effectively pan nuclear so a segmented analysis
123 Figure 4-7 – Adaptations of the assay: Comet, Alamar Blue and Inhibitors (a) 48br cells were treated 8x2min fractions of alpha radiation at 0.8 Gy/min and analyzed by alkaline comet assay; data from two independent experiments. Representative images of comet tails at three time points (data contributed by ND Berger). (b) Alamar blue performed on 48br cells 24hr after treatment. Cell were exposed to 2-16x 2min fractions of alpha radiation at 0.8 Gy/min (red) , mock treated(air exposed) cells treat with 8 or 16 fractions blue), cells treated with 1 μM staurosporine was included as a positive control (black). The error bars indicate the SD observed between 3 different replicates. (c) Mean nuclear ɣ-H2AX intensity of matched IF analysis of cells treated as in panel at 3 h:(a). (d) 48br cells pretreated with indicated inhibitors and irradiated with 2min alpha IR and fixed after 1hr. Average volume of ɣ-H2AX objects per cell nucleus as percentage of the nuclear volume charted by condition. did not produce meaningful data. To solve this issue, I used the nuclear segmentation as a 124 contour for γH2AX quantification, which is then quantified as the total nuclear γH2AX signal. This quantification approach (of matched wells irradiated at same time) shows a γH2AX signal peak at 15 min post IR of the final dose. Please note that that is ~3 hrs after commencing the initial IR fraction, and the effect of this timing is apparent in the 0 h which has significant γH2AX signalling as would be expected if the 2-4hr peak with high LET IR is durable. What is clear that the γH2AX signal is reduced at 1 hr post IR, and that the repair is still proceeding in these cells, showing that these doses did not preclude repair and did not universally activate apoptosis, although a 40% loss in viability was evident (fig 4-7 b/c). Any use of the assay based on IF staining that uses DAPI can also be sued to interrogate the cell cycle (appendix B).The system can also be used (appendix B) this has also bee previously demonstrated but future adds to the adaptability of the setup223. Thus, with some nuances, my system can report on cell viability. In the future, it will be interesting to explore whether cells irradiated using this approach can be recovered, counted and plated within the context of a clonogenic survival assay. This would require a tight cell seeding modality, so no cells were outside the mapped irradiation field (fig 4-1a). However, this technical barrier should be surmountable, as I have already demonstrated this exact process within the context of the alkaline comet assay discussed in the previous section.
4.2.5 Screening potential and adaptability to other genetic model systems
To demonstrate this system’s usefulness in compound/drug screening systems, a small selection of small molecules known to impact γH2AX signalling via PIKK inhibition were investigated. Pre-treatment of cells with inhibitors of the ATM and DNA-PK protein kinases (referred to as ATMi and DPKi, respectively), or a combination of both, prior to 0.8 Gy high LET IR significantly reduced γH2AX signal. ATMi treatment alone lead to a 35% reduction in overall γH2AX signal, while DPKi lead to a 32% reduction. The combination lead to a >99% reduction in signal, to that point that there was no observable significant difference when compared the unirradiated cells (fig4-7d). These results are important as they reveal strong contributions from both ATM and DNA-PK to the γH2AX signalling after high LET IR, somewhat distinct to the smaller role understood to be played by DNA-PK during low LET IR induced DSB signalling. Interestingly, there
125 seems to be a variation in the degree of contribution of each enzyme to the overall signal in differing cell lines; however, I did not assess the roots of this phenomena within the context of this thesis and speculate it may represent differing levels of the respective protein kinase activities in different cell types. Genetic model organisms have driven much of our understanding of DNA damage and repair biology for the past century. Both budding and fission yeast models have made foundational contributions to the field of radiation biology but have seen very little work on high LET radiation. Indeed, a PubMed search of the terms “yeast” and “radiation” together reveal close to 8,500 articles, of which just over 4,000 have been produced since the dawn of the 21st century. Of these, only 30 articles across the entire recorded literature examine high LET IR, of which only 9 are from the 21st century. This indicates a major paucity of information on high LET IR biology produced using one of humanities most powerful genetic systems. To open this area for further investigation, I also developed an adapted procedure for the irradiation of Saccharomyces cerevisiae (Baker’s yeast), one of the most commonly investigated types of yeast within modern biology. The procedure is operably comparable to the mammalian tissue culture methodology, where cells are grown on a flat surface an exposed to high LET IR through the positioning of the 241Am source at a controlled distance from the cells for a given length of time (appendix D). Yeast are notoriously resistant to IR, and typical experiments with low LET IR utilize doses in the range of 100 Gy or more to observe effects. Thus, I decided to grow yeast cells under the 241Am sources for 4, 8 16 & 24 hr before removing the IR source and allowing cells time to recover. One major advantage in this case being that yeast grow naturally in air, and do not require liquid media when plated on the surface of rich media agar plates. Thus, sham controls were dispensable when working with yeast and this system. The IR exposures for the chosen periods of exposure equate to chronic doses at 0.4 Gy/min totaling 96, 192, 384 and 576 Gy, respectively (appendix D). Plates of yeast were seeded in different dilutions in a colony format and were provided by Dr. Jennifer Cobb’s laboratory. Using this approach, I demonstrate a dose dependent loss of colony forming ability across a 100-600 Gy dose range in wildtype cells. I also was able to examine a canonically IR-sensitive yeast mutant, namely one genetically ablated for the key DSB sensor protein Rad50 (an
126 essential part of the MRN complex) and observed the ‘expected’ increase in comparative sensitivity for the RAD50 mutant strain. Indeed, yeast lacking Rad50 showed a clear and greater loss in colony forming potential at the 192 Gy and 384 Gy doses. Given the right tools, my new high LET IR system will enable critical new investigation into the effects of high LET radiation in yeast systems, and the important take-home message from my work is that these are now demonstrably possible in standard lab conditions.
4.3 Discussion
In this study, I developed and validated a benchtop, 96 well plate-based method for alpha particle irradiation in living cells. I have shown that alpha particle irradiation induces a robust DNA damage response in multiple eukaryotic cell types that is quantifiable with a 3D, TANGO-based computer-assisted image analysis pipeline. Alpha particle irradiation produces DNA damage that is amenable to quantification through physical assays like the alkaline comet assay and has a significant effect on cell viability at relatively low doses. The results of this system are consistent with previous literature exploring the effects of alpha particle irradiation in a biological context. Previous studies have utilized many different approaches to alpha particle irradiation. These include calibrated but low- throughput 241Am pin-point sources, only able to irradiate a small number of cells grown on delicate Mylar film 213,224–227 or alpha particle beams from particle accelerators in an adjacent facility 228. It is worth mentioning that I and my colleagues in the lab originally explored the utility of Mylar-based cultures for irradiation. Whilst highly transformed cell lines such as Hela cells were able to grow on these thin sheets of plastic, normal human fibroblast cultures and even A549 human lung adenocarcinoma cells all underwent rapid cell death and culture failure. The reasons for this were not apparent but precluded the use of that methodology for many applications. I speculate that a functional p53 axis and propensity towards programmed cell death may preclude use of Mylar film as a physical base for human cell cultures, as this highly flexible surface provides an unstable surface and, at least in our hands, observable fluctuations in shearing and fluid dynamic stresses on the cells229. Mylar is also enormously fragile, and for adapting assays for any kind of medium to high throughput format, its use is prohibitive.
127 Previous studies found, overall, that alpha particles induce pan-nuclear γH2AX phosphorylation dependent upon ATM219, and that maximal γH2AX foci appear over 1 hour post alpha particle irradiation and take longer to resolve than foci from low LET radiation 218. Unsurprisingly, work with alpha particle point sources also revealed that high LET radiation produces more chromosomal rearrangements than γ-rays, and induces more robust apoptosis than an equivalent dose of γ-rays 224,225,230. In this study, I have shown that as little as 1.6 Gy of alpha particle radiation results in measurable decreases in cell viability 32 hours after exposure, and doses up to 12.8 Gy lead to approximately 40% reductions in cell viability. I have also shown that 6.4 Gy of alpha particle irradiation induces DNA damage quantifiable by alkaline comet assay that repairs in a time- dependent manner. My setup represents significant improvements and innovation over a single manufactured 241Am point source setup. When compared to the use of a single manufactured source setup 213, our 3D printed 96 well plate setup provides much more scale, speed and adaptability, which could facilitate broader investigation into this understudied area with considerably higher throughput potential. In terms of cost, a single manufactured point source 241Am has costed between $500-8000 (depending on vendor and year reported), while the cost of a single smoke detector (from which 241Am dioxide pellets are extracted) is approximately $25 each before tax. While a significant advantage of our setup is its medium-to-high throughput nature, a 96 well plate irradiation setup with our system would only cost an estimated $2880, based on $30 smoke detectors. Smoke detectors are available on a consumer level, and with access to a 3D printer, this setup is easy to replicate. I should stress, however, that permission to disassemble smoke detectors must be obtained from the specific atomic / nuclear regulator for a given jurisdiction, as it is illegal for a non-specialist to do this due to associated radiation hazards. I obtained permission from Atomic Energy of Canada, a Canadian crown corporation that regulates applications for such devices. Standard operating protocols for smoke detector disassembly and source extraction were developed in conjunction with local radiation safety officers, and 241Am sources were then logged and stored in secure locations when not in use. The relative amounts of radiation for each 241Am source is actually negligible (kBq) in comparison with most research or industrial radioactivity
128 sources (MBq-TBq), and thus storage and safety concerns are minimised; however, it is important to follow appropriate safety protocols. With technical advantages aside, this technique is also compatible with a multitude of experimental setups, which will allow for both broad hypotheses and high- throughput investigations. One potential avenue for investigation is the examination of galactic cosmic radiation (GCR) on biological systems, in the context of deep space exploration. GCR exposure is one of the most relevant dangers of extended space exploration, and has been shown to increase cancer risk, induce cognitive dysfunction and increase the prevalence of dementia 231–233. GCR is comprised of energetic protons and 12% alpha particles, and thus my system can contribute to a ground-based research effort to mimic and study the effects of GCR exposure on cells and tissues. As the culture vessels are optical 96 well plates, our method is also highly compatible with current super-resolution microscopy techniques and could be combined with techniques to monitor large-scale processes like chromatin remodeling. Many unanswered questions concerning the relative contributions of ATM and DNA-PK to the alpha particle-induced DNA damage response are amenable to examination in a system like this. Altogether, the potential impacts of this method, combined with its feasibility, accessibility and adaptability make it an ideal setup to interrogate the cellular effects of high LET radiation.
129 Chapter Five: Interpretation
5.1 Main Findings
My data shows that residential radon is indeed an issue in Alberta and Saskatchewan, and that a considerable percentage of the population are being exposed to this radioactive gas at home. Through geographic mapping, I also determined that certain regions display radon levels that are, on average, higher than other regions, particularly the electoral districts of the Regina area. At a more individual house level, I confirm that there is considerable variability in radon levels, that there are no uniformly ‘safe’ areas within the survey region, and that the range of values is considerable, yet follows the known pattern of a long positive tail in the distribution. I found that, for homes reporting year of construction, the newer homes trended higher in radon readings. I also showed that specific home metrics and occupant behaviours correlate with differing indoor air radon concentrations. Resulting from my interest in the biological effects of high LET radiation exposure, I also developed a method to investigate human cell exposure using 241Am. The energy profile of 241Am alpha emissions closely matches that of decaying radon, and I successfully designed prototyped and implemented a setup that uses 241Am pellets to irradiate human cells in a 96wp format. I showed that the setup was capable of demonstrating DSB induction signaling by radiation using immunofluorescence γH2AX quantification. I also showed that the imaging and analysis phases can be successfully automated. This setup is robust in many regards, capable of dealing with multiple cell types, adaptable to multiple assays, including viability assays, and well designed for screening protocols.
5.2 Explanation and Comparison
Residential radon has been previously documented in Canada, and readings have been shown to vary geographically176. The patterns of geographic distribution I observed (i.e. Regina averaging higher) matches some the previously reported data. However, in an overall consideration, our testing shows generally higher radon levels than has previously been reported by Health Canada across the survey region. The reason for prior lower
130 reporting is unclear, as the testing protocols I used are based on the Health Canada recommendations. It is possible that there was an effect of under reporting in previous work, as my samples size is significantly higher than that of the Cross-Canada study. While it was unsurprising and has been widely reported that specific metrics and occupant behaviours might correlate to radon concentrations, it is important that such patterns are determined locally, as the interaction of geology climate and build practice all vary geographically. The finding that newer constructions correlated with higher radon levels was novel and had not been previously reported in a Canadian context. If such a trend were to continue, then the expectation would be increased population exposure to alpha particle radiation with time, as more and more newer builds enter the housing stock. The physical alpha particle irradiation system that I designed and developed to model such exposures performed well in our hands, and was capable of reproducing some of the known idiosyncratic effects of high LET radiation. Additionally, it was highly adaptable, being compatible with a 96 well plate that gives the system a very broad base for other assay protocols that could be plugged in. I show that this setup easily operates with immunofluorescence and fluorescence protocols, reading out on microscopes or plate readers. Compared to other methods of studying radon exposure at the cell biology level, this system is much more accessible to most labs than particle accelerator or radon chamber-based methods. Directly comparing to a previously published methods, I find that this system is capable of all that thosr previous methods can achieve (i.e. single cell analysis) but operates with greater throughput. Of note, the previous methodologies did not require media removal for irradiation, but did require culturing cell on a custom culture surface of custom-built mylar plastic-covered bottom dishes, as opposed to the glass bottom 96wp we used which are widely available from multiple vendors. My method thus circumvents the difficulty of achieving a taut mylar plastic surface necessary for successful cell culture, and avoids the frustration linked to breaking the otherwise very fragile and often brittle mylar plastic surface.
5.3 Strengths and limitations
The high level of engagement is perhaps the greatest strength of the radon mapping studies, and I observed that there was a large amount of public engagement, which is an
131 asset to a program such as this, where the power for avoiding the risk largely lies with individual home owners. The high engagement from the general public in the citizen science aspects of my work also lead to high numbers of homes testing, which translated to larger sample sizes and facilitated more precise measurements. These strengths come from the fact that the studies were open, with convenience-based participant selection that enabled high rates of testing. At the same time, this may also be considered a limitation, as although as it meant that I did not draw randomly from the population. That is to say those who engaged with the study may have been somewhat skewed demographically, and this could be checked by comparing things the build type to the overall housing stock in an area using municipal datasets or by issuing a demographic survey in parallel to the home metric survey. It remains important that such studies are carried out in a local setting, as any findings that link home metrics and occupant behaviours to indoor radon levels are likely to be influenced by local weather patterns and build practices. It was beyond the scope of this work to begin to question these secondary links, but one might well imagine that window opening behaviours, and how that might affect indoor radon, would be heavily influenced by outdoor air temperature and broader meteorological trends for a given area. Additionally, in my initial study, I showed that sub-slab depressurization was an effective intervention in the Greater Calgary area, but the local building code demands that a layer of loose clean gravel exists below the slab. As the presence of this intentionally facilities air movement across the area under the slab, this better enables radon mitigation but may not be the standard practice elsewhere. Having large mapped radon datasets will allow future investigators to begin asking these more specific questions using real time radon data coupled to weather data (Appendix E), or by correlating trends with major changes in legislation covering building practices.
In terms of the specific strength of the irradiation system, certainly compared to other methods for studying high LET radiation is has considerable practicality and adaptability, both of which have been mentioned earlier. The only specialised equipment required to perform this is access to a 3D printer, which are increasingly common in research institutes and universities. The 241Am pellets themselves must be procured, and
132 local regulatory issues must be considered (i.e. it is generally illegal for a member of the lay-public to disassemble a smoke detector without special permission); however, the 241Am pellets have a relatively low risk from a health and safety perspective, and a much lower cost than the conceivable alternatives. All other equipment is standard to most laboratories and widely available. One concern about the protocol is the need for media removal during irradiation. I see this as an unavoidable but relatively minor risk, and whilst it is not required in the previously published protocol, that mylar-film based cell culture system is less user- friendly, less scalable, and (at least in my case) I and my colleagues could never reproduce it regularly in our lab. Top down irradiation with a high LET radiation of a comparable energy level to the alpha emissions from decaying radon will never penetrate any normal amount of aqueous tissue culture media, and so not only is this is why the 241Am pellets are relatively safe to handle, but also why radon correlates mostly with lung cancer, as the alpha particle radiation cannot penetrate the layer of dead skin above the living fibroblasts in the epidermis. To be absorbed into living tissue, it has to be inhaled or internalized. I address this concern and, for the cells I use in the specific assays outlined within this thesis, I ran matched air-exposed controls that compared favorably to the completely untreated controls that remained in media. I suggest that these relevant controls are always performed using the method I developed, especially with new users, different cell lines or adapting it for other assays. However, at the same time, I would expect the standard 2 min air exposure time will perform well in most settings and not represent a major experimental caveat. A second concern for this irradiation system is the dose rate, its magnitude and variability. The low dose rate could be considered more a feature than a ‘bug’, as it far better represents the modality of typical human exposure to high LET radiation – which is low dose and chronic via radon inhalation. Although, if it was seen as a significant barrier to a specific experiment, one would only need to source pellets with high concentration of 241Am in the alloy to deliver more radiation over a given amount of time. The variability between sources is something that could also be addressed by engaging with a manufacturer; however, I find the variability compares favourably with that of other typical radiation systems (137Cs, GammaCell), and it is straightforward to control for this
133 by always internally duplicating conditions within experiments using two different sources. The individual pellets dose rate didn’t vary much between repeats or over time and, considering that the dose variability in reality is easy to quantify, one could easily remove significantly outlying pellets if desired.
5.4 Consequences
The implications of the radon campaign I helped to design and execute are evident in the sample numbers. Indeed, it is a major public health priority (in preventing radon induced cancers) to get homes tested, and I have achieved considerable work in this regard. It also vital that the testing is carried out in parallel to an education program and accessible public awareness campaign, since it is important to give people the information required to make decisions to test or mitigate as appropriate. The broad uptake in the testing has shown us that our efforts meet some success. Another major aim is to inform the policy making agenda, and here I am thrilled to have also seen some success. Indeed, in fall of 2017, Bill 209 “The Alberta Radon Awareness and Testing Act” was introduced at the Alberta legislator by Member of Legislative Assembly (MLA) Robyn Luff. Bill 209 was passed after three readings by unanimous consent, and the author of this bill acknowledges that she drew inspiration from my work and the radon testing campaign (Appendix G). The study (now called the “Evict Radon campaign”) is ongoing and expanding and, based on my findings, is asking further questions that will be developed as the hypotheses refine. Within the context of a cell biology lab, as others begin to use the alpha particle irradiation protocol, work aimed at identifying the genetic susceptibility factors or potential genetic signature for and of radon induced lung cancers will take shape. Taking this multi-disciplinary approach has been fruitful, and I hope that my findings and protocols will take the research in to a setting that might inform longitudinal epidemiology, molecular characterisation and clinical studies. As an aside, I also believe that the investigation of high LET radiation has potential as a ‘game-changing’ avenue of discovery science, and hope the system and protocols detailed here can be used to ask
134 some of the many unanswered and indeed unasked questions about high LET radiobiology. A target for all public health work concerned with cancer is to reduce the burden through preventative measures. I showed that radon exposure in Alberta and Saskatchewan is substantially higher than previously reported, and that newer constructions correlate with higher radon levels. Might this mean that radon induced lung cancers are more prevalent that previous estimates, and that newer buildings will only add to the issue? As always, there are more experiments to run and questions to ask, but I believe the data supports these concerns. I also show that there are effective interventions to reduce the exposure. It remains important that people test for residential radon, as it is a pressing issue of public health, but it is a problem with known solutions. It is important to neither overstate nor understate the risk.
135
136 Chapter Six: Materials and Methods
6.1 Mammalian Cell Culture Work
6.1.1 Cell lines and Tissue Culture
HeLa and HEK-293 (ATCC© CRL-1573) cells were grown in DMEM (Gibco) supplemented with 10 % (v/v) FCS (Hyclone), 1 % (v/v) Pen-Strep. 1BR-hTERT and 48BR primary human fibroblast cells were grown in DMEM (Gibco) supplemented with 15 % (v/v) FCS (Hyclone), 1 % (v/v) Pen-Strep. Cells were grown at 37°C, 5% CO2 humidified incubators.
6.1.2 Transfections
To transfected plasmids or siRNA into HEK293 and HeLa cells they were passaged to ~70% confluency the day prior to transfection for 2 4hr timepoints. All transfections were carried out using jetPRIME (POLYPLUS 114-01) exactly according to manufacturer instructions.
6.2 Reagents
6.2.1 Antibodies
Target Company Species dilute IF/IB 53BP1 Abcam, ab21083 Rabbit 1:800 IF Actin Abcam, ab3280 Mouse 1:3000 IB CHD3 Abcam, ab84528 Rabbit 1:500 IB CHD4 Abcam, ab54603 Mouse 1:500 IB CHD6 Abcam, ab51330 Mouse 1:200 IB CHD6 LSBio, LS-C342418 Rabbit 1:50 IF GFP Santa-Cruz, sc996 Mouse 1:1000 IB KAP-1 Abcam, ab10584 Rabbit 1:2000 IB Trevigen, 4336-PBC- PAR Rabbit 1:400 IF 100
137 γH2AX Abcam, ab26350 Mouse 1:800 IF Anti-mouse Thermo Fisher Goat 1:400 IF Alexa488 Scientific Anti-mouse Thermo Fisher Goat 1:400 IF Alexa594 Scientific Anti-rabbit Thermo Fisher Goat 1:400 IF Alexa488 Scientific Anti-rabbit Thermo Fisher Goat 1:400 IF Alexa594 Scientific
6.2.2 Inhibitors & Compounds
Reagent Manufacturer
Phenylmethylsulfonyl Fluoride (PMSF) Sigma Wortmannin (WM) Sigma Microcystin-LR (MC-LR) Sigma AZD2281/Olaparib (PARP inhibitor) Selleck Chemicals NU7441 (KU57788, DNA-PK inhibitor) Selleck Chemicals KU-55933 (ATM inhibitor) Tocris Bromodeoxyuridine (BrdU) BD Biosciences 6.2.3 Solutions & Buffers
1 X PBS 137 mmol/L NaCL, 10 mmol/L Na2HPO4, 2.7 mmol KCl, and 1.8 mmol/L KH2PO4 NETN Buffer 0.15 M NaCl, 0.25 mM EDTA, 50 mM Tris-HCl [pH 8.0], and 1% (v/v) NP-40 Sample Buffer 80 mM Tris-HCl pH 6.8/ 2% SDS/ 5% 2-mercaptoethanol/ 10% (v/v) glycerol/ 0.1% (w/v) bromophenol blue Electroblot Buffer 48 mM Tris/ 39 mM glycine/ 20% (v/v) ethanol
138 6.3 Alpha/Gamma IR and 241Am Source Dosimetry
For IF based methods, 3.0E+3 to 5.0E+3 cells were seeded in 200uL per well 24 hr before irradiation. The day of irradiation, the 241Am sources were placed into the scaffolds, and media was temporarily removed from the wells of interest. 241Am sources were lowered over the cells and allowed to irradiate wells for 2 minutes (or as indicated) before being removed, and the media replaced. The 2 min irradiation was standard to deliver a 1.6 Gy dose; matched air exposure (sham) controls and untreated controls were incorporated for each experiment. Gamma ray exposed samples were irradiated using a GammaCell 1000 Elite source, which contains a 137Cs source that emits approximately 2.9 Gy/min. For Alamar Blue viability and comet assays, 1.0E+3 cells were plated in 5 μL droplets positioned in the center of each well, were allowed to adhere for 8 hours, and then wells were carefully topped up with 195 μL media. For irradiation, a maximum 2- minute fractionation regimen was performed, to minimize the time cells were exposed to air outside of media. Cells were irradiated for 2 minutes, the media replaced, and they were returned to the incubator for 10 min intervals between fractions. This was repeated for the desired number of fractions, and matched air exposure controls were incorporated for each experiment. EBT3 GafChromic film (Ashland Specialty Ingredients, Wayne, NJ) was used to quantify the dose rate of 8 independent 241Am sources. EBT3 GafChromic film is sensitive from 0.1 cGy to upwards of 10 Gy, has sub-millimeter spatial resolution, develops in real time with radiation dose (i.e. does not require any chemical processing or fixation), and has near tissue-equivalence. For these reasons, it is frequently employed to perform absolute dosimetry in the context of radiation therapy234–236. EBT3 film is composed of a 28 μm radiation sensitive layer sandwiched between two 125 μm matte polyester layers. The top polyester layer was removed for the purposes of this study, as the path length of the alpha particles emitted by 241Am (~35 μm in polyester and ~43 μm in water237) is too short to penetrate beyond the surface layer of the film. The split film was calibrated with a high energy linear accelerator (Varian TrueBeam LINAC [Varian, Palo Alto, CA]) using a 6MV photon beam from 0 to 6 Gy. All films were scanned with an Epson Expression 12000XL flatbed scanner (Seiko Epson Corp., Nagano, Japan). The film calibration followed the methods outlined by Morrison et al.238 The scanned optical
139 densities were converted to dose using a Triple-channel heterogeneity correction method, utilizing all three color channels (Red, green, and blue) from the scanned image to correct for film thickness heterogeneities.238 To achieve the appropriate dose level, the split EBT3 film was exposed by the 241Am sources for 6-10 minutes, achieving doses within 2-5 Gy. Three exposures were performed for each source with different time intervals which were used to calculate a dose rate for each source in Gy/min. The uncertainty in the determined dose rates was at most 3.6% with a coverage factor k=2.
6.4 Design and 3D Printing
All 3D designs were created in SketchUp Free 2017 (Trimble), designs were converted to the .stl file format and prepared for 3D printing in Ultimaker Cura 3.5. Printing was carried out on the Ultimaker2 platform using PLA plastic.
6.5 Immunofluorescence
Coverslips were washed in 99% (v/v) ethanol, then distilled water before being autoclaved. Cells were grown on square or circle coverslips to ~70% confluency. Following specific cell treatment, coverslips were washed briefly with PBS and fixed at RT in PBS containing 3% (w/v) paraformaldehyde and 2% (w/v) sucrose for 10 minutes. Cells were then washed three with 5 min washes in 1X PBS before being permeabilized with 0.2% (v/v) Triton X-100 at RT for 10 minutes. Cells were then washed briefly with 1X PBS and blocked at RT for 30 minutes with 1X PBS containing 2% (w/v) fat-free dried milk powder. Primary and secondary antibodies were diluted in PBS solution containing 1% (w/v) BSA, and a total volume of 100 μL was placed onto coverslips for the desired incubation period. For immunofluorescence in a 96 well plate, cells were fixed with 3% (w/v) PFA in 1X PBS at the indicated time points at room temperature for 10 minutes. Working gently with a multichannel pipette, cells were washed three times with 1X PBS containing 0.05% (v/v) Tween 20 (PBST), and permeabilized in 1X PBS containing 0.25% (v/v) Trixon X-100 for 5 minutes at room temperature. Cells were washed three times with PBST, before being blocked in 3% (v/v) normal goat serum (NGS) in PBST for 10 minutes at room temperature. Cells were then incubated with primary antibodies in 1% NGS in PBST for 1 hour at room temperature and washed three times in PBST. Cells
140 were then incubated with secondary antibodies in 1% (v/v) NGS/PBST for 25 minutes at room temperature, washed three times with PBST, and counterstained with 1X PBS containing DAPI (1:10,000). Wells were then filled with 1X PBS for imaging.
6.6 Widefield Epifluorescence Microscope Imaging
Samples were imaged with or scored on a Zeiss Axio Observer Z1 platform microscope, with a Plan-Apochromat 20x/0.8, an EC Plan-Neofluar 40x/0.75 or a Plan-Apochromatin 63x/1.4 objective and an AxioCam MRm Rev.3 camera. Acquisition software Zen Pro (Zeiss), for the automated acquisition of the 96 well plate images the Z-stack, autofocus and position modules of Zen must be available to program the acquisition.
6.7 Image Analysis
Zen Blue, ImageJ and Tango I used for image analysis. Images were all acquired on Zeiss systems and saved as cvi. files. These files were imported into ImageJ using the Bioformats plugin. Within TANGO, the default setting was used for Nuclear edge detection. This was applied to the imported .czi images to pull out individual nuclei from fields, tiles and positions contained in the experiments. Incorrectly segmented nuclei were manually removed. To analyze the H2AX signal, an initial sliding paraboloid 2D background subtraction was applied to each Z-stack. Spot detection used the Renyi Entropy method to automatically detect and segment H2AX signal within the nuclei. Post filters were applied to remove any signal found outside the nuclear segmentation or totaling less than 2-pixel area. Once the nuclear and H2AX segmentation had run measurements were exported based on mapping these voxel volumes to the original data.
6.8 Confocal Microscope Imaging
Samples were imaged with LSM880 Carl Zeiss Airyscan confocal microscope, with a Plan Apochromat 20×/0.8 NA, an EC Plan Neofluar 40×/0.75 NA, or a Plan Apochromat 63×/1.4 NA objective and a camera (AxioCam MRm Rev.3; Carl Zeiss) or GaAsP or Airyscan detectors (Carl Zeiss). Acquisition and analysis software used was Zen Blue and Zen Black (Carl Zeiss).
141 6.9 Statistical Analysis
Statistical analysis was carried out using Excel (16.0.6769.2017, Microsoft), GraphPad Prism (7.0, GraphPad), SPSS (24, IBM), RStudio (1.1.456, RStudio) and R (version 3.4.4, R Core Team). Specific tests and data handling is described in the relevant chapters.
6.10 Alkaline comet assay
To assess alpha particle induced DNA damage, the alkaline comet assay was performed as per Caldecott et al. 2009 239, with some modifications. Briefly, center plated cells were irradiated with 8 fractions of α particle IR or exposed to air under the “sham” conditions, with 8 wells per condition used. Cells were trypsinized and collected and resuspended in pre-chilled 1X PBS. The cells were spun down at 3000 RPM for 3 minutes and resuspended in 150 μL of cold 1X PBS. The volume of cells was mixed with an equal volume of 1.2% (w/v) low melting point agarose, and quickly layered onto a pre-chilled fully frosted glass slide with a 0.8% (w/v) agarose base layer. Coverslips were used to maintain the agarose shape, and slides were chilled at 4˚C in the dark until set. Coverslips were removed, and slides were immersed in pre-chilled lysis buffer (2.5M NaCl, 10 mM Tris-HCl, 100 mM EDTA, 1% (v/v) Triton X-100, 10% (v/v) DMSO, pH 10) for 1 h in the dark, then washed three times with pre-chilled electrophoresis buffer (50 mM NaOH, 1 mM EDTA, 10% (v/v) DMSO). Slides were placed in electrophoresis chamber with buffer, and allowed to equilibrate for 45 minutes before being electrophoresed at 25 V for 25 minutes (0.6 V/cm), and neutralized with 1ml Tris-HCl pH 7.4 for 1 h at 4˚C. DNA was stained with SYBR Green (1:10,000) and 0.5% antifade (0.5% (w/v) p- phenylenediamine in 20 mM Tris HCl, pH 8.8, 90 % (v/v) glycerol) for 10 minutes at room temperature. Slides were imaged and Comet Assay IV (Perceptive Instruments, UK) software was used to quantify tail moment for at least 100 cells per experiment.
6.11 Alamar Blue cell viability assay
Examination of cell viability after α particle IR was performed with the Alamar blue assay. Cells were center plated in a 96 well plate, and exposed to 2, 4, 8 or 16 fractions of α particle IR, or matching sham controls. 1 μM staurosporine was included as a positive control to induced cell death / loss of viability (Sigma, S5921). 2-minute fractions were
142 separated with 5 minute media rest periods. 24 hours after α particle IR, Alamar blue was added to a final concentration of 10% (w/v), and cells were allowed to metabolize the dye over 8 hours. Fluorescence was then read on a Molecular Devices SpectraMax iD3 plate reader, with excitation at 560 nm and emission at 600 nm. A fluorescence correction was performed to empty media to determine percent dye reduced, and data were normalized to untreated, unirradiated cells.
6.12 Cell Lysate Preparations and Western Blotting
Cell extracts were prepared and lysed in NETN buffer with phosphatase and protease inhibitors as described above. Lysis buffer was added to cells, and cells were left to sit on ice for 30 min. Samples were then sonicated and centrifuged at 10,000 x g for 10 minutes at 4°C. Protein concentrations were then determined using BSA standard curve, followed by boiling for 3 minutes in 4x SB. Samples were then run on SDS-PAGE gels containing 10% (w/v) acrylamide (resolving gel) and 5% (w/v) acrylamide (stacking gel) unless otherwise noted. Electrophoresis was run at 60V for the initial 10min and subsequently were run at 120V until the loading dye reached the end of the gel. Protein samples were transferred to nitrocellulose membrane at 100 volts for 60 minutes at RT using electroblot buffer unless otherwise stated. After primary antibody incubation, a secondary antibody was used against appropriate host. After washing, nitrocellulose blots were developed using Enhanced Chemiluminescence (ECL) solution and subsequently developed using x- ray film (Fuji).
6.12.1 Immunoprecipitation.
For examining interactions in HEK293 cells, extracts were prepared by lysing cells in 1X PBS + 1% (v/v) NETN buffer, as described in western blot procedure. For IP by of GFP tagged proteins, 10 μL slurry of GFP-trap beads (ChromoTek) were equilibrated in 500 μL of 1X PBS + 1% (v/v) NETN buffer. Beads were then loaded with 500 μg of protein extract for 2 hrs with rocking at 4°C. Beads were next washed in 1X PBS + 1% (v/v) NETN buffer three times. Samples were boiled in 40 μL of 6X sample buffer at 95°C and were then ready for SDS-PAGE.
143 6.13 UV Laser Microirradiation and Live Cell Imaging
Cells of interest were cultured onto round microscope coverslips (Fisherbrand, 12-545- 102 25CIR.-1) in 3 cm culture dishes overnight, prior to transfection. Cells were then transiently transfected (as described above) with 2 μg DNA 24 hours prior to irradiation. BrdU (10 μM) was added 24 hrs prior to irradiation. Coverslips containing transfected and BrdU-labelled cells were placed into a metal round cassette suitable for live cell imaging. Cells were washed in 1X PBS once and then 1 mL of 1X PBS was added to cover cells in metal cassette for live cell imaging. When inhibitors were used they were added to the culture media 1 hr prior to the irradiation and maintained in the PBS during imaging. For live cell imaging, coverslips in the metal cassette were placed into a pre- warmed incubation chamber on the 355nm UV-Laser PALM MicroBeam system. Cells were irradiated using 35% laser power and cut at 10 um/s. Five images were taken before irradiation to serve as pre-damage images and after laser cutting images were acquire at 1 min intervals for a 21 min unless otherwise indicated.
144 References
1. Rontgen, W. C. On A New Kind of Ray, a Prelimianry Communication. Wurzbg. Physico-Medical Soc. (1895). 2. Bécquerel, A. H. Sur les radiations invisibles emises par les corps phosphorescents. Comptes Rendus Math. Acad. des Sci. Paris 122, 501 (1896). 3. Bécquerel, A. H. Sur les radiations invisibles émises par les sels d’uranium. Comptes Rendus Math. Acad. des Sci. Paris 122, 689–694 (1896). 4. Curie, M. Rayons émis par les composes de l’uranium et du thorium. Comptes Rendus l’Académie des Sci. Paris 126, 1101–1103 (1898). 5. Rutherford, E. Uranium radiation and the electrical conduction produced by it. Philosophical Magazine Series 5 47, (1899). 6. Rutherford, E. The magnetic and electric deviation of the easily absorbed rays from radium. Philos. Mag. Ser. 6 5, 177–187 (1903). 7. Debierne, A. Sur l’émanation de l’actinium. Comptes Rendus l’Académie des Sci. Paris 138, 411–414 (1904). 8. Rutherford, E. & Soddy, F. XLI. The cause and nature of radioactivity .—Part I. London, Edinburgh, Dublin Philos. Mag. J. Sci. 4, 370–396 (1902). 9. Rutherford, E. & Soddy, F. LXIV. The cause and nature of radioactivity .—Part II. London, Edinburgh, Dublin Philos. Mag. J. Sci. 4, 569–585 (1902). 10. Rutherford, E. & Andrade, E. N. da C. XXXI. The spectrum of the penetrating γ rays from radium B and radium C. London, Edinburgh, Dublin Philos. Mag. J. Sci. 28, 263–273 (1914). 11. Bertolotti, M. The Discovery: Victor F. Hess and the Balloon Ascents. in 33–44 (2013). doi:10.1007/978-3-642-28371-0_2 12. Millikan, R. A. & Bowen, I. S. High Frequency Rays of Cosmic Origin I. Sounding Balloon Observations at Extreme Altitudes. Phys. Rev. 27, 353–361 (1926). 13. Alvarez, L. & Compton, A. H. A Positively Charged Component of Cosmic Rays. Phys. Rev. 43, 835–836 (1933). 14. Wilson, C. T. R. On an Expansion Apparatus for Making Visible the Tracks of
145 Ionising Particles in Gases and Some Results Obtained by Its Use. Proc. R. Soc. A Math. Phys. Eng. Sci. 87, 277–292 (1912). 15. Wilson, C. T. R. On a Method of Making Visible the Paths of Ionising Particles through a Gas. Proc. R. Soc. A Math. Phys. Eng. Sci. 85, 285–288 (1911). 16. Thomson, J. J. & McClelland, J. A. JJ Thomson and JA McClelland. in Proc. Cambridge Phil. Soc Vol. 9 (1896). 17. Thomson, J. J. XIX. On the theory of the conduction of electricity through gases by charged ions. London, Edinburgh, Dublin Philos. Mag. J. Sci. 47, 253–268 (1899). 18. Sommerfeld, A. Uber Die Beugung der Rontgenstrahlen. Ann. Phys. 343, 473–506 (1912). 19. Bragg, S. W. H. The Structure of Organic Crystals. Proc. Phys. Soc. London 34, 33–50 (1921). 20. Bragg, W. H. & Bragg, W. L. The Reflection of X-rays by Crystals. Proc. R. Soc. A Math. Phys. Eng. Sci. 88, 428–438 (1913). 21. Auterhoff, H. Nobel Lectures Physiology or Medicine 1901–1921. XVI, 5615. Elsevier Publishing Company, Amsterdam-London-New York 1967. Preis: £ 9.0.0. Preis des Satzes von 3 Bänden: £ 24.0.0. Arch. Pharm. (Weinheim). 301, 310–310 (1968). 22. ASTBURY, W. T. & BELL, F. O. X-Ray Study of Thymonucleic Acid. Nature 141, 747–748 (1938). 23. Watson, J. D. & Crick, F. H. Molecular structure of nucleic acids: a structure for deoxyribose nucleic acid. Am. J. Psychiatry 160, 623–4 (2003). 24. Muller, H. J. The Problem of Genetic Modification. Proc. 5th Int. Congr. 1, 234– 260 (1927). 25. Muller, H. J. ARTIFICIAL TRANSMUTATION OF THE GENE. Science 66, 84– 7 (1927). 26. ICRP. ICRP 103: The 2007 Recommendations of the International Commission on Radiological Protection. Ann. ICRP 37, 330 (2007). 27. Who Handbook indoor radon: a public health perspective. (WHO Press, 2009). 28. Society, M. S. M. Maryland Medical Journal, Vol. 57: July, 1914 (Classic Reprint). (Forgotten Books, 1914).
146 29. Zeubline, E. The Present Status of Radioactive Therapy in Medicine. Marylanf Med. J. 57, (1915). 30. Copeland, R. S. To prohibit the movement in interstate commerce of adulterated and misbranded food, drugs, devices, and cosmetics, and for other purposes. 75– 717 (75th United States Congress, 1938). 31. Miller, C. E., Hasterlik, R. J. & Finkel, A. J. The argonne radium studies. Summary of fundamental data. ANL-7531 and ACRH-106. ANL Rep. 1–596 (1968). 32. Evans, R. D. Radium poisoning II: The quantitative determination of the radium content and radium elimination rate of living persons. Am J Roentgenol Radium Ther Nucl Med 37, 368–378 (1937). 33. Evans, R. D. & Aub, J. C. Recent Progress in the Study of Radium Poisoning. Am. Assoc. Adv. Sci. 4, 227–233 (1937). 34. Preston, D. L., Shimizu, Y., Pierce, D. A., Suyama, A. & Mabuchi, K. Studies of mortality of atomic bomb survivors. Report 13: Solid cancer and noncancer disease mortality: 1950-1997. Radiat. Res. 160, 381–407 (2003). 35. Krewski, D. et al. Projection of residential radon lung cancer risks: the Beir VI risk models. Radiat. Prot. Dosimetry 102, 371-3; author reply 373-4 (2002). 36. Health Risks from Exposure to Low Levels of Ionizing Radiation. (National Academies Press, 2006). doi:10.17226/11340 37. Hueper, W. C. Occupational Tumors and Allied Diseases. (Charles C. Thomas., 1942). 38. Hueper, W. C. & United States Public Health Service. Environmental and occupational cancer. (U.S. Government printing Office, 1948). 39. HUEPER, W. C. Silicosis, asbestosis, and cancer of the lung. Am. J. Clin. Pathol. 25, 1388–90 (1955). 40. Harting, F. H. & Hesse, W. Der lungenkrebs, die Bergkrankheit in den Schneeberger gruben. Vjschr. Gerichtl. Med. Offentl. Gesundheitswes. 31, 102– 132,313–337 (1979). 41. Lorenz, E. Radioactivity and Lung Cancer; A Critical Review of Lung Cancer in the Miners of Schneeberg and Joachimsthal. JNCI J. Natl. Cancer Inst. 5, 1–15
147 (1944). 42. Lundin, F. J., Wagoner, J. & Archer, V. Radon daughter exposure and respiratory cancer quantitative and temporal aspects. (1971). 43. International Agency for Research on Cancer. IARC Working Group on the Evaluation of Carcinogenic Risk to Humans. (IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, 2012). 44. Health Effects of Exposure to Radon. (National Academies Press, 1999). doi:10.17226/5499 45. Brugge, D. & Goble, R. The history of uranium mining and the Navajo people. Am. J. Public Health 92, 1410–9 (2002). 46. Kesminiene, A. et al. Risk of hematological malignancies among Chernobyl liquidators. Radiat. Res. 170, 721–735 (2008). 47. Rahu, K. et al. Chernobyl cleanup workers from Estonia: Cohort description and related epidemiological research. J. Radiol. Prot. 35, R35–R45 (2015). 48. Bennett, B., Repacholi, M. & Carr, Z. Health Effects of the Chernobyl Accident and Special Health Care Programmes. Reports UN Chernobyl Forum 1–160 (2006). 49. Hanahan, D. & Weinberg, R. a. Hallmarks of cancer: the next generation. Cell 144, 646–674 (2011). 50. Canadian Cancer Statistics Advisory Committee. Canadian Cancer Statistics. Toronto, Can. Cancer Soc. (2018). 51. Lindahl, T. & Wood, R. D. Quality control by DNA repair. Science 286, 1897–905 (1999). 52. Storchova, Z. & Pellman, D. From polyploidy to aneuploidy, genome instability and cancer. Nat. Rev. Mol. Cell Biol. 5, 45–54 (2004). 53. Lindahl, T. Instability and decay of the primary structure of DNA. Nature 362, 709–15 (1993). 54. Boiteux, S. & Guillet, M. Abasic sites in DNA: repair and biological consequences in Saccharomyces cerevisiae. DNA Repair (Amst). 3, 1–12 (2004). 55. Hoeijmakers, J. H. J. DNA damage, aging, and cancer. N. Engl. J. Med. 361, 1475– 85 (2009).
148 56. Harper, J. W. & Elledge, S. J. The DNA damage response: ten years after. Mol. Cell 28, 739–45 (2007). 57. Caldecott, K. W. Single-strand break repair and genetic disease. Nat. Rev. Genet. 9, 619–31 (2008). 58. Flynn, R. L. & Zou, L. ATR: a master conductor of cellular responses to DNA replication stress. Trends Biochem. Sci. 36, 133–140 (2011). 59. Ciccia, A. & Elledge, S. J. The DNA damage response: making it safe to play with knives. Mol. Cell 40, 179–204 (2010). 60. Bekker-Jensen, S. & Mailand, N. Assembly and function of DNA double-strand break repair foci in mammalian cells. DNA Repair (Amst). 9, 1219–1228 (2010). 61. Bekker-Jensen, S. et al. Spatial organization of the mammalian genome surveillance machinery in response to DNA strand breaks. J. Cell Biol. 173, 195– 206 (2006). 62. Matsuoka, S. et al. ATM and ATR substrate analysis reveals extensive protein networks responsive to DNA damage. Science 316, 1160–6 (2007). 63. Stanley, F. K. T., Moore, S. & Goodarzi, A. A. CHD chromatin remodelling enzymes and the DNA damage response. Mutat. Res. 750, 31–44 (2013). 64. Hiom, K. Coping with DNA double strand breaks. DNA Repair (Amst). 9, 1256– 1263 (2010). 65. Kim, J.-S. et al. Independent and sequential recruitment of NHEJ and HR factors to DNA damage sites in mammalian cells. J. Cell Biol. 170, 341–7 (2005). 66. Abraham, R. T. PI 3-kinase related kinases: ‘big’ players in stress-induced signaling pathways. DNA Repair (Amst). 3, 883–887 (2004). 67. Kim, J.-A., Kruhlak, M., Dotiwala, F., Nussenzweig, A. & Haber, J. E. Heterochromatin is refractory to gamma-H2AX modification in yeast and mammals. J. Cell Biol. 178, 209–218 (2007). 68. Riballo, E. et al. A pathway of double-strand break rejoining dependent upon ATM, Artemis, and proteins locating to gamma-H2AX foci. Mol. Cell 16, 715–724 (2004). 69. Shibata, A. et al. Role of ATM and the damage response mediator proteins 53BP1 and MDC1 in the maintenance of G(2)/M checkpoint arrest. Mol. Cell. Biol. 30,
149 3371–3383 (2010). 70. Bergink, S. & Jentsch, S. Principles of ubiquitin and SUMO modifications in DNA repair. Nature 458, 461–7 (2009). 71. Luijsterburg, M. S. & van Attikum, H. Chromatin and the DNA damage response: the cancer connection. Mol. Oncol. 5, 349–67 (2011). 72. Noon, A. T. & Goodarzi, A. a. 53BP1-mediated DNA double strand break repair: Insert bad pun here. DNA Repair (Amst). 10, 1071–1076 (2011). 73. Chapman, J. R. & Jackson, S. P. Phospho-dependent interactions between NBS1 and MDC1 mediate chromatin retention of the MRN complex at sites of DNA damage. EMBO Rep. 9, 795–801 (2008). 74. Spycher, C. et al. Constitutive phosphorylation of MDC1 physically links the MRE11–RAD50–NBS1 complex to damaged chromatin. J. Cell Biol. 181, 227– 240 (2008). 75. Melander, F. et al. Phosphorylation of SDT repeats in the MDC1 N terminus triggers retention of NBS1 at the DNA damage–modified chromatin. J. Cell Biol. 181, 213–226 (2008). 76. Noon, A. T. et al. 53BP1-dependent robust localized KAP-1 phosphorylation is essential for heterochromatic DNA double-strand break repair. Nat. Cell Biol. 12, 177–184 (2010). 77. Lee, J.-H., Goodarzi, A. a, Jeggo, P. a & Paull, T. T. 53BP1 promotes ATM activity through direct interactions with the MRN complex. EMBO J. 29, 574–585 (2010). 78. Poulsen, M., Lukas, C., Lukas, J., Bekker-Jensen, S. & Mailand, N. Human RNF169 is a negative regulator of the ubiquitin-dependent response to DNA double-strand breaks. J. Cell Biol. 197, 189–199 (2012). 79. Chen, J., Feng, W., Jiang, J., Deng, Y. & Huen, M. S. Y. Ring Finger Protein RNF169 Antagonizes the Ubiquitin-dependent Signaling Cascade at Sites of DNA Damage. J. Biol. Chem. 287, 27715–27722 (2012). 80. Danielsen, J. R. et al. DNA damage–inducible SUMOylation of HERC2 promotes RNF8 binding via a novel SUMO-binding Zinc finger. J. Cell Biol. 197, 179–187 (2012).
150 81. Bekker-Jensen, S. et al. HERC2 coordinates ubiquitin-dependent assembly of DNA repair factors on damaged chromosomes. Nat. Cell Biol. 12, 80-6; sup pp 1- 12 (2010). 82. Galanty, Y., Belotserkovskaya, R., Coates, J. & Jackson, S. P. RNF4, a SUMO- targeted ubiquitin E3 ligase, promotes DNA double-strand break repair. Genes Dev. 26, 1179–1195 (2012). 83. Yin, Y. et al. SUMO-targeted ubiquitin E3 ligase RNF4 is required for the response of human cells to DNA damage. Genes Dev. 26, 1196–1208 (2012). 84. Galanty, Y. et al. Mammalian SUMO E3-ligases PIAS1 and PIAS4 promote responses to DNA double-strand breaks. Nature 462, 935–9 (2009). 85. Chapman, J. R. et al. RIF1 is essential for 53BP1-dependent nonhomologous end joining and suppression of DNA double-strand break resection. Mol. Cell 49, 858– 71 (2013). 86. Di Virgilio, M. et al. Rif1 Prevents Resection of DNA Breaks and Promotes Immunoglobulin Class Switching. Science (80-. ). 339, 711–715 (2013). 87. Zimmermann, M., Lottersberger, F., Buonomo, S. B., Sfeir, A. & de Lange, T. 53BP1 Regulates DSB Repair Using Rif1 to Control 5’ End Resection. Science (80-. ). 339, 700–704 (2013). 88. Huen, M. S. Y. et al. Regulation of Chromatin Architecture by the PWWP Domain-Containing DNA Damage-Responsive Factor EXPAND1/MUM1. Mol. Cell 37, 854–864 (2010). 89. Goodarzi, A. a & Jeggo, P. a. Irradiation induced foci (IRIF) as a biomarker for radiosensitivity. Mutat. Res. (2011). doi:10.1016/j.mrfmmm.2011.05.017 90. Dobbs, T. A., Tainer, J. A. & Lees-Miller, S. P. A structural model for regulation of NHEJ by DNA-PKcs autophosphorylation. DNA Repair (Amst). 9, 1307–1314 (2010). 91. Zhang, X. et al. Artemis is a phosphorylation target of ATM and ATR and is involved in the G2/M DNA damage checkpoint response. Mol. Cell. Biol. 24, 9207–20 (2004). 92. Wang, M. et al. PARP-1 and Ku compete for repair of DNA double strand breaks by distinct NHEJ pathways. Nucleic Acids Res. 34, 6170–6182 (2006).
151 93. Haince, J.-F. et al. Ataxia telangiectasia mutated (ATM) signaling network is modulated by a novel poly(ADP-ribose)-dependent pathway in the early response to DNA-damaging agents. J. Biol. Chem. 282, 16441–53 (2007). 94. Haince, J.-F. et al. PARP1-dependent kinetics of recruitment of MRE11 and NBS1 proteins to multiple DNA damage sites. J. Biol. Chem. 283, 1197–1208 (2008). 95. Williams, G. J., Lees-Miller, S. P. & Tainer, J. A. Mre11-Rad50-Nbs1 conformations and the control of sensing, signaling, and effector responses at DNA double-strand breaks. DNA Repair (Amst). 9, 1299–1306 (2010). 96. López-Contreras, A. J. & Fernandez-Capetillo, O. The ATR barrier to replication- born DNA damage. DNA Repair (Amst). 9, 1249–1255 (2010). 97. Burrows, A. E. & Elledge, S. J. How ATR turns on: TopBP1 goes on ATRIP with ATR. Genes Dev. 22, 1416–21 (2008). 98. Humphreys, T. D. & Wold, M. S. Eukaryotic Replication Protein A. in Encylopedia of Life Sciences 1–11 (John Wiley & Sons, Ltd: Chichester, 2010). doi:10.1002/9780470015902.a0001051.pub2 99. Goodarzi, A. a, Jeggo, P. & Lobrich, M. The influence of heterochromatin on DNA double strand break repair: Getting the strong, silent type to relax. DNA Repair (Amst). 9, 1273–82 (2010). 100. Symington, L. L. S. & Gautier, J. Double-strand break end resection and repair pathway choice. Annu. Rev. Genet. 45, 247–71 (2011). 101. Goodarzi, A. A. & Jeggo, P. A. The Heterochromatic Barrier to DNA Double Strand Break Repair: How to Get the Entry Visa. Int. J. Mol. Sci. 13, 11844–11860 (2012). 102. Shibata, A., Conrad, S., Birraux, J. & Geuting, V. Factors determining DNA double-strand break repair pathway choice in G2 phase. EMBO … 30, 1079–92 (2011). 103. Beucher, A. et al. ATM and Artemis promote homologous recombination of radiation-induced DNA double-strand breaks in G2. EMBO J. 28, 3413–3427 (2009). 104. Goodarzi, A. a et al. ATM signaling facilitates repair of DNA double-strand breaks associated with heterochromatin. Mol. Cell 31, 167–77 (2008).
152 105. Woodbine, L., Brunton, H., Goodarzi, a a, Shibata, A. & Jeggo, P. a. Endogenously induced DNA double strand breaks arise in heterochromatic DNA regions and require ataxia telangiectasia mutated and Artemis for their repair. Nucleic Acids Res. 39, 6986–97 (2011). 106. Ziv, Y. et al. Chromatin relaxation in response to DNA double-strand breaks is modulated by a novel ATM- and KAP-1 dependent pathway. Nat. Cell Biol. 8, 870–6 (2006). 107. Jakob, B. et al. DNA double-strand breaks in heterochromatin elicit fast repair protein recruitment, histone H2AX phosphorylation and relocation to euchromatin. Nucleic Acids Res. 39, 6489–99 (2011). 108. Cowell, I. G. et al. gammaH2AX foci form preferentially in euchromatin after ionising-radiation. PLoS One 2, e1057 (2007). 109. Falk, M., Lukásová, E. & Kozubek, S. Chromatin structure influences the sensitivity of DNA to gamma-radiation. Biochim. Biophys. Acta 1783, 2398–414 (2008). 110. Schuster-Böckler, B. & Lehner, B. Chromatin organization is a major influence on regional mutation rates in human cancer cells. Nature 488, 504–507 (2012). 111. HADA, M. & GEORGAKILAS, A. G. Formation of Clustered DNA Damage after High-LET Irradiation: A Review. J. Radiat. Res. 49, 203–210 (2008). 112. Anderson, R. M., Stevens, D. L. & Goodhead, D. T. M-FISH analysis shows that complex chromosome aberrations induced by -particle tracks are cumulative products of localized rearrangements. Proc. Natl. Acad. Sci. 99, 12167–12172 (2002). 113. Eccles, L. J., O’Neill, P. & Lomax, M. E. Delayed repair of radiation induced clustered DNA damage: Friend or foe? Mutat. Res. Mol. Mech. Mutagen. 711, 134–141 (2011). 114. GOODHEAD, D. T. Mechanislns for the Biological Effectiveness of High-LET Radiations. J. Radiat. Res. 40, 1–13 (1999). 115. Ward, J. F. Biochemistry of DNA Lesions. Radiat. Res. 104, S103 (1985). 116. Goodhead, D. T. Initial Events in the Cellular Effects of Ionizing Radiations: Clustered Damage in DNA. Int. J. Radiat. Biol. 65, 7–17 (1994).
153 117. Ward, J. F. The Complexity of DNA Damage: Relevance to Biological Consequences. Int. J. Radiat. Biol. 66, 427–432 (1994). 118. Bennett, P. V, Cintron, N. S., Gros, L., Laval, J. & Sutherland, B. M. Are endogenous clustered dna damages induced in human cells? Free Radic. Biol. Med. 37, 488–499 (2004). 119. Semenenko, V. A. & Stewart, R. D. A Fast Monte Carlo Algorithm to Simulate the Spectrum of DNA Damages Formed by Ionizing Radiation. Radiat. Res. 161, 451– 457 (2004). 120. Nikjoo, H., O’Neill, P., Terrissol, M. & Goodhead, D. T. Quantitative modelling of DNA damage using Monte Carlo track structure method. Radiat. Environ. Biophys. 38, 31–38 (1999). 121. Datta, K., Neumann, R. D. & Winters, T. A. Characterization of complex apurinic/apyrimidinic-site clustering associated with an authentic site-specific radiation-induced DNA double-strand break. Proc. Natl. Acad. Sci. 102, 10569– 10574 (2005). 122. Dextraze, M.-E., Gantchev, T., Girouard, S. & Hunting, D. DNA interstrand cross- links induced by ionizing radiation: An unsung lesion. Mutat. Res. Mutat. Res. 704, 101–107 (2010). 123. Jenner, T. J., DeLara, C. M., O’Neill, P. & Stevens, D. L. Induction and Rejoining of DNA Double-strand Breaks in V79-4 Mammalian Cells Following γ- and α- irradiation. Int. J. Radiat. Biol. 64, 265–273 (1993). 124. LOBRICH, P. K. COOPER and B. RYDBER, M. Non-random distribution of DNA double-strand breaks induced by particle irradiation. Int. J. Radiat. Biol. 70, 493–503 (1996). 125. Stenerlöw, B., Höglund, E., Carlsson, J. & Blomquist, E. Rejoining of DNA fragments produced by radiations of different linear energy transfer. Int. J. Radiat. Biol. 76, 549–57 (2000). 126. Kühne, M., Rothkamm, K. & Löbrich, M. No dose-dependence of DNA double- strand break misrejoining following alpha-particle irradiation. Int. J. Radiat. Biol. 76, 891–900 (2000). 127. Anderson, J. A., Harper, J. V., Cucinotta, F. A. & O’Neill, P. Participation of
154 DNA-PKcs in DSB Repair after Exposure to High- and Low-LET Radiation. Radiat. Res. 174, 195–205 (2010). 128. Alloni, D. et al. A Monte Carlo Study of the Radiation Quality Dependence of DNA Fragmentation Spectra. Radiat. Res. 173, 263–271 (2010). 129. Campa, A. et al. DNA Fragmentation Induced in Human Fibroblasts by 56 Fe Ions: Experimental Data and Monte Carlo Simulations. Radiat. Res. 171, 438–445 (2009). 130. Stenerlöw, B., Höglund, E., Elmroth, K., Karlsson, K. H. & Radulescu, I. Radiation quality dependence of DNA damage induction. Radiat. Prot. Dosimetry 99, 137–41 (2002). 131. PANG, D. et al. Radiation-generated Short DNA Fragments May Perturb Non- homologous End-joining and Induce Genomic Instability. J. Radiat. Res. 52, 309– 319 (2011). 132. Mukherjee, K. & Storici, F. A Mechanism of Gene Amplification Driven by Small DNA Fragments. PLoS Genet. 8, e1003119 (2012). 133. Pearson, C. G. Enhanced mutagenic potential of 8-oxo-7,8-dihydroguanine when present within a clustered DNA damage site. Nucleic Acids Res. 32, 263–270 (2004). 134. Malyarchuk, S. Two clustered 8-oxo-7,8-dihydroguanine (8-oxodG) lesions increase the point mutation frequency of 8-oxodG, but do not result in double strand breaks or deletions in Escherichia coli. Nucleic Acids Res. 32, 5721–5731 (2004). 135. GRIFFIN, C. . et al. Effectiveness of 0.28 keV carbon K ultrasoft X-rays at producing simple and complex chromosome exchanges in human fibroblasts in vitro detected using FISH. Int. J. Radiat. Biol. 73, 591–598 (1998). 136. Schuster-Böckler, B. & Lehner, B. Chromatin organization is a major influence on regional mutation rates in human cancer cells. Nature 488, 504–7 (2012). 137. Smith, K. S., Liu, L. L., Ganesan, S., Michor, F. & De, S. Nuclear topology modulates the mutational landscapes of cancer genomes. Nat. Struct. Mol. Biol. 24, 1000–1006 (2017). 138. Goodarzi, A. a, Kurka, T. & Jeggo, P. a. KAP-1 phosphorylation regulates CHD3
155 nucleosome remodeling during the DNA double-strand break response. Nat. Struct. Mol. Biol. 18, 831–839 (2011). 139. Asaithamby, A. et al. Repair of HZE-Particle-Induced DNA Double-Strand Breaks in Normal Human Fibroblasts. Radiat. Res. 169, 437–446 (2008). 140. Tobias, F., Durante, M., Taucher-Scholz, G. & Jakob, B. Spatiotemporal analysis of DNA repair using charged particle radiation. Mutat. Res. Mutat. Res. 704, 54–60 (2010). 141. Leatherbarrow, E. L., Harper, J. V., Cucinotta, F. A. & O’Neill, P. Induction and quantification of γ-H2AX foci following low and high LET-irradiation. Int. J. Radiat. Biol. 82, 111–118 (2006). 142. Aten, J. A. Dynamics of DNA Double-Strand Breaks Revealed by Clustering of Damaged Chromosome Domains. Science (80-. ). 303, 92–95 (2004). 143. Desai, N. et al. Immunofluorescence Detection of Clustered γ-H2AX Foci Induced by HZE-Particle Radiation. Radiat. Res. 164, 518–522 (2005). 144. Jakob, B., Scholz, M. & Taucher-Scholz, G. Biological imaging of heavy charged- particle tracks. Radiat. Res. 159, 676–84 (2003). 145. Nakajima, N. I. et al. Visualisation of γH2AX Foci Caused by Heavy Ion Particle Traversal; Distinction between Core Track versus Non-Track Damage. PLoS One 8, e70107 (2013). 146. Meyer, B. et al. Clustered DNA damage induces pan-nuclear H2AX phosphorylation mediated by ATM and DNA–PK. Nucleic Acids Res. 41, 6109– 6118 (2013). 147. Radulescu, I., Elmroth, K. & Stenerlöw, B. Chromatin Organization Contributes to Non-randomly Distributed Double-Strand Breaks after Exposure to High-LET Radiation. Radiat. Res. 161, 1–8 (2004). 148. Okayasu, R. et al. Repair of DNA Damage Induced by Accelerated Heavy Ions in Mammalian Cells Proficient and Deficient in the Non-homologous End-Joining Pathway. Radiat. Res. 165, 59–67 (2006). 149. Singleton, B. K. et al. Molecular and biochemical characterization of xrs mutants defective in Ku80. Mol. Cell. Biol. 17, 1264–1273 (1997). 150. Errami, A. et al. Ku86 defines the genetic defect and restores X-ray resistance and
156 V(D)J recombination to complementation group 5 hamster cell mutants. Mol. Cell. Biol. 16, 1519–1526 (1996). 151. Chen, F., Peterson, S. R., Story, M. D. & Chen, D. J. Disruption of DNA-PK in Ku80 mutant xrs-6 and the implications in DNA double-strand break repair. Mutat. Res. Repair 362, 9–19 (1996). 152. Wang, H., Wang, X., Zhang, P. & Wang, Y. The Ku-dependent non-homologous end-joining but not other repair pathway is inhibited by high linear energy transfer ionizing radiation. DNA Repair (Amst). 7, 725–733 (2008). 153. Arosio, D., Costantini, S., Kong, Y. & Vindigni, A. Fluorescence Anisotropy Studies on the Ku-DNA Interaction. J. Biol. Chem. 279, 42826–42835 (2004). 154. Wang, H. et al. Characteristics of DNA-binding proteins determine the biological sensitivity to high-linear energy transfer radiation. Nucleic Acids Res. 38, 3245– 3251 (2010). 155. Lücke-Huhle, C. Similarities between human ataxia fibroblasts and murine SCID cells: high sensitivity to gamma rays and high frequency of methotrexate-induced DHFR gene amplification, but normal radiosensitivity to densely ionizing alpha particles. Radiat. Environ. Biophys. 33, 201–10 (1994). 156. Rodier, F. et al. DNA-SCARS: distinct nuclear structures that sustain damage- induced senescence growth arrest and inflammatory cytokine secretion. J. Cell Sci. 124, 68–81 (2011). 157. Oliveira, N. G. et al. Wortmannin enhances the induction of micronuclei by low and high LET radiation. Mutagenesis 18, 37–44 (2003). 158. Ritter, S. & Durante, M. Heavy-ion induced chromosomal aberrations: a review. Mutat. Res. 701, 38–46 (2010). 159. Darby, S. et al. Radon in homes and risk of lung cancer: Collaborative analysis of individual data from 13 European case-control studies. Br. Med. J. 330, 223–226 (2005). 160. Schauer, D. A. & Linton, O. W. NCRP Report No. 160, Ionizing Radiation Exposure of the Population of the United States, medical exposure--are we doing less with more, and is there a role for health physicists? Health Phys. 97, 1–5 (2009).
157 161. Henshaw, D. L., Eatough, J. P. & Richardson, R. B. Radon as a causative factor in induction of myeloid leukaemia and other cancers. Lancet (London, England) 335, 1008–12 (1990). 162. Marsh, J. W. & Bailey, M. R. A review of lung-to-blood absorption rates for radon progeny. Radiat. Prot. Dosimetry 157, 499–514 (2013). 163. Teras, L. R. et al. Residential radon exposure and risk of incident hematologic malignancies in the Cancer Prevention Study-II Nutrition Cohort. Environ. Res. 148, 46–54 (2016). 164. Olsen, A. H., Parkin, D. M. & Sasieni, P. Cancer mortality in the United Kingdom: projections to the year 2025. Br. J. Cancer 99, 1549–54 (2008). 165. Collier, C. G. et al. Carcinogenicity of radon/radon decay product inhalation in rats – effect of dose, dose rate and unattached fraction. Int. J. Radiat. Biol. 81, 631–647 (2005). 166. Darby, S., Hill, D. & Doll, R. Radon: A likely carcinogen at all exposures. Ann. Oncol. 12, 1341–1351 (2001). 167. Li, B.-Y. & Tong, J. Adverse Effects Attributed to Long-term Radon Inhalation in Rats. J. Toxicol. Environ. Heal. Part A 70, 925–930 (2007). 168. Nie, J.-H. et al. Oxidative Damage in Various Tissues of Rats Exposed to Radon. J. Toxicol. Environ. Heal. Part A 75, 694–699 (2012). 169. Hendry, J. H. et al. Human exposure to high natural background radiation: what can it teach us about radiation risks? J. Radiol. Prot. 29, A29–A42 (2009). 170. Darby, S. et al. Residential radon and lung cancer--detailed results of a collaborative analysis of individual data on 7148 persons with lung cancer and 14,208 persons without lung cancer from 13 epidemiologic studies in Europe. Scand. J. Work. Environ. Health 32 Suppl 1, 1–83 (2006). 171. Field, R. W. et al. An overview of the North American residential radon and lung cancer case-control studies. J. Toxicol. Environ. Heal. - Part A 69, 599–631 (2006). 172. Lubin, J. H. & Boice, J. D. Lung Cancer Risk From Residential Radon: Meta- analysis of Eight Epidemiologic Studies. JNCI J. Natl. Cancer Inst. 89, 49–57 (1997).
158 173. Parkin, D. M., Bray, F., Ferlay, J. & Pisani, P. Global Cancer Statistics, 2002. CA. Cancer J. Clin. 55, 74–108 (2005). 174. Thun, M. J. et al. Lung cancer death rates in lifelong nonsmokers. J. Natl. Cancer Inst. 98, 691–9 (2006). 175. Sun, S., Schiller, J. H. & Gazdar, A. F. Lung cancer in never smokers--a different disease. Nat. Rev. Cancer 7, 778–90 (2007). 176. Health Canada. Cross-Canada Survey of Radon Concentrations in Homes Final Report. 1–29 (2012). 177. Chen, J., Moir, D. & Whyte, J. Canadian population risk of radon induced lung cancer: a re-assessment based on the recent cross-Canada radon survey. Radiat. Prot. Dosimetry 152, 9–13 (2012). 178. Canadian Cancer Society’s Advisory Committee on Cancer Statistics. Canadian Cancer Statistics 2016. (2016). 179. Centers for Disease Control and Prevention (CDC). Smoking-attributable mortality, years of potential life lost, and productivity losses--United States, 2000- 2004. MMWR. Morb. Mortal. Wkly. Rep. 57, 1226–8 (2008). 180. Zhou, W. et al. Polymorphisms in the DNA repair genes XRCC1 and ERCC2, smoking, and lung cancer risk. Cancer Epidemiol. Biomarkers Prev. 12, 359–65 (2003). 181. Lubin, J. H. et al. Lung cancer in radon-exposed miners and estimation of risk from indoor exposure. J. Natl. Cancer Inst. 87, 817–27 (1995). 182. Krewski, D. et al. Residential radon and risk of lung cancer: a combined analysis of 7 North American case-control studies. Epidemiology 16, 137–45 (2005). 183. Lewis, B. J. et al. Cosmic Radiation Exposure on Canadian-Based Commercial Airline Routes. Radiat. Prot. Dosimetry 86, 7–24 (1999). 184. Sigurdson, A. J. & Ron, E. Cosmic Radiation Exposure and Cancer Risk Among Flight Crew. Cancer Invest. 22, 743–761 (2004). 185. Armato, S. G., Doshi, D. J., Engelmann, R., Caligiuri, P. & MacMahon, H. Temporal subtraction of dual-energy chest radiographs. Med. Phys. 33, 1911–1919 (2006). 186. Simpson, J. A. Elemental and Isotopic Composition of the Galactic Cosmic Rays.
159 Annu. Rev. Nucl. Part. Sci. 33, 323–382 (1983). 187. Cucinotta, F., Kim, M. & Chappell, L. Space radiation cancer risk projections and uncertainties-2010. NASA Tech. Pap., NASA/TP- … (2011). 188. Zeitlin, C. et al. Measurements of energetic particle radiation in transit to Mars on the Mars Science Laboratory. Science 340, 1080–4 (2013). 189. Castro, J. R. et al. Treatment of cancer with heavy charged particles. Int. J. Radiat. Oncol. 8, 2191–2198 (1982). 190. Antonovic, L., Brahme, A., Furusawa, Y. & Toma-Dasu, I. Radiobiological description of the LET dependence of the cell survival of oxic and anoxic cells irradiated by carbon ions. J. Radiat. Res. 54, 18–26 (2013). 191. TSUBOI, K. et al. Cell Cycle Checkpoint and Apoptosis Induction in Glioblastoma Cells and Fibroblasts Irradiated with Carbon Beam. J. Radiat. Res. 48, 317–325 (2007). 192. HAMADA, N. et al. Recent Advances in the Biology of Heavy-Ion Cancer Therapy. J. Radiat. Res. 51, 365–383 (2010). 193. Stanley, F. K. T. et al. Comprehensive survey of household radon gas levels and risk factors in southern Alberta. C. open 5, E255–E264 (2017). 194. Luff, R. Radon Awareness and Testing Act. 1–8 (Legislative Assembly of Alberta, 2017). 195. Edwards, A. A., Purrott, R. J., Prosser, J. S. & Lloyd, D. C. The induction of chromosome aberrations in human lymphocytes by alpha-radiation. Int. J. Radiat. Biol. Relat. Stud. Phys. Chem. Med. 38, 83–91 (1980). 196. Little, J. B., Kennedy, A. R. & McGandy, R. B. Effect of dose rate on the induction of experimental lung cancer in hamsters by alpha radiation. Radiat. Res. 103, 293– 9 (1985). 197. Moore, S., Stanley, F. K. T. & Goodarzi, A. A. The repair of environmentally relevant DNA double strand breaks caused by high linear energy transfer irradiation--no simple task. DNA Repair (Amst). 17, 64–73 (2014). 198. Chen, J. Canadian lung cancer relative risk from radon exposure for short periods in childhood compared to a lifetime. Int. J. Environ. Res. Public Health 10, 1916– 26 (2013).
160 199. Canadian Cancer Society. Canadian Cancer Statistics Special topic : Predictions of the future burden of cancer in Canada. Public Heal. Agency Canada 1–151 (2015). doi:Canadian Cancer Society 200. Chen, J. A review of radon doses. Radiat. Prot. Manag. 22, 27–31 (2005). 201. Milner, J. et al. Home energy efficiency and radon related risk of lung cancer: modelling study. BMJ 348, f7493–f7493 (2014). 202. ACI Committee 209. Guide for modeling and calculating shrinkage and creep in hardened concrete. (2008). 203. Gilbert, R. I. Shrinkage, Cracking and Deflection- the Serviceability of Concrete Structures. Electron. J. Struct. Eng. 1, 2–14 (2001). 204. Statistics Canada & Canadian Institute for Health Information. Canadian Health Indicators. 2016002 (2016). doi:82-221-X2016002 205. Cancer, I. A. for R. on. Agents Classified by the IARC Monographs. (2015). 206. Barros, N. G., Steck, D. J. & Field, R. W. A Comparison of Winter Short-term and Annual Average Radon Measurements in Basements of a Radon-prone Region and Evaluation of Further Radon Testing Indicators. Health Phys. 106, 535–544 (2014). 207. Vaupotič, J. & Kobal, I. Correlation between Short-Term and Long-Term Radon Measurements. Isotopes Environ. Health Stud. 38, 39–46 (2002). 208. Al-Jarallah, M. I., Fazal-ur-Rehman & Abdalla, K. Comparative study of short- and long-term indoor radon measurements. Radiat. Meas. 43, S471–S474 (2008). 209. Fontana, A. O. et al. Differential DNA repair pathway choice in cancer cells after proton- and photon-irradiation. Radiother. Oncol. 116, 374–380 (2015). 210. Timm, S., Lorat, Y., Jakob, B., Taucher-Scholz, G. & Rübe, C. E. Clustered DNA damage concentrated in particle trajectories causes persistent large-scale rearrangements in chromatin architecture. Radiother. Oncol. (2018). doi:10.1016/j.radonc.2018.07.003 211. Lorat, Y., Timm, S., Jakob, B., Taucher-Scholz, G. & Rübe, C. E. Clustered double-strand breaks in heterochromatin perturb DNA repair after high linear energy transfer irradiation. Radiother. Oncol. 121, 154–161 (2016). 212. Carter, R. J. et al. Complex DNA Damage Induced by High Linear Energy
161 Transfer Alpha-Particles and Protons Triggers a Specific Cellular DNA Damage Response. Int. J. Radiat. Oncol. 100, 776–784 (2018). 213. Stap, J. et al. Induction of linear tracks of DNA double-strand breaks by α-particle irradiation of cells. Nat. Methods 5, 261–266 (2008). 214. Löbrich, M. et al. γH2AX foci analysis for monitoring DNA double-strand break repair. 9, 662–669 (2010). 215. Gudla, P. R. et al. A high-throughput system for segmenting nuclei using multiscale techniques. Cytom. Part A 73, 451–466 (2008). 216. Ollion, J., Cochennec, J., Loll, F., Escudé, C. & Boudier, T. TANGO: A generic tool for high-throughput 3D image analysis for studying nuclear organization. Bioinformatics 29, 1840–1841 (2013). 217. Löbrich, M. et al. gammaH2AX foci analysis for monitoring DNA double-strand break repair: strengths, limitations and optimization. Cell Cycle 9, 662–9 (2010). 218. Staaf, E., Brehwens, K., Haghdoost, S., Czub, J. & Wojcik, A. Gamma-H2AX foci in cells exposed to a mixed beam of X-rays and alpha particles. Genome Integr. 3, 8 (2012). 219. Horn, S., Brady, D. & Prise, K. Alpha particles induce pan-nuclear phosphorylation of H2AX in primary human lymphocytes mediated through ATM. Biochim. Biophys. Acta - Mol. Cell Res. 1853, 2199–2206 (2015). 220. Olive, P. L. & Banáth, J. P. The comet assay: a method to measure DNA damage in individual cells. Nat. Protoc. 1, 23–29 (2006). 221. Collins, A. R. The Comet Assay for DNA Damage and Repair: Principles, Applications, and Limitations. Mol. Biotechnol. 26, 249–261 (2004). 222. Olive, P. L. & Banáth, J. P. The comet assay: a method to measure DNA damage in individual cells. Nat. Protoc. 1, 23–9 (2006). 223. Roukos, V., Pegoraro, G., Voss, T. C. & Misteli, T. Cell cycle staging of individual cells by fluorescence microscopy. Nat. Protoc. 10, 334–48 (2015). 224. Franken, N. A. P. et al. Comparison of RBE values of high-LET α-particles for the induction of DNA-DSBs, chromosome aberrations and cell reproductive death. Radiat. Oncol. 6, 64 (2011). 225. Chauhan, V., Howland, M., Chen, J., Kutzner, B. & Wilkins, R. C. Differential
162 effects of alpha-particle radiation and x-irradiation on genes associated with apoptosis. Radiol. Res. Pract. 2011, 679806 (2011). 226. Munshi, N. C. et al. Identification of genes modulated in multiple myeloma using genetically identical twin samples. Blood 103, 1799–806 (2004). 227. Chauhan, V., Howland, M., Chen, J., Kutzner, B. & Wilkins, R. C. Differential Effects of Alpha-Particle Radiation and X-Irradiation on Genes Associated with Apoptosis. Radiol. Res. Pract. 2011, 1–9 (2011). 228. Hei, T. K. et al. Mutagenic effects of a single and an exact number of particles in mammalian cells. Proc. Natl. Acad. Sci. 94, 3765–3770 (1997). 229. Chen, C. S. Geometric Control of Cell Life and Death. Science (80-. ). 276, 1425– 1428 (1997). 230. Chaudhary, P. et al. Variations in the Processing of DNA Double-Strand Breaks Along 60-MeV Therapeutic Proton Beams. Int. J. Radiat. Oncol. 95, 86–94 (2016). 231. Parihar, V. K. et al. Cosmic radiation exposure and persistent cognitive dysfunction. Sci. Rep. 6, 34774 (2016). 232. Cucinotta, F. A., Alp, M., Sulzman, F. M. & Wang, M. Space radiation risks to the central nervous system. Life Sci. Sp. Res. 2, 54–69 (2014). 233. Chancellor, J., Scott, G. & Sutton, J. Space Radiation: The Number One Risk to Astronaut Health beyond Low Earth Orbit. Life 4, 491–510 (2014). 234. Devic, S. Radiochromic film dosimetry: past, present, and future. Phys. Med. 27, 122–34 (2011). 235. Reinhardt, S., Hillbrand, M., Wilkens, J. J. & Assmann, W. Comparison of Gafchromic EBT2 and EBT3 films for clinical photon and proton beams. Med. Phys. 39, 5257–62 (2012). 236. Lewis, D., Micke, A., Yu, X. & Chan, M. F. An efficient protocol for radiochromic film dosimetry combining calibration and measurement in a single scan. Med. Phys. 39, 6339–50 (2012). 237. Berger, M. J. ESTAR, PSTAR, and ASTAR: Computer programs for calculating stopping-power and range tables for electrons, protons, and helium ions. NIST Interagency/Internal Rep. (1992). 238. Morrison, H., Menon, G. & Sloboda, R. S. Radiochromic film calibration for low-
163 energy seed brachytherapy dose measurement. Med. Phys. 41, 072101 (2014). 239. Cortes Ledesma, F., El Khamisy, S. F., Zuma, M. C., Osborn, K. & Caldecott, K. W. A human 5’-tyrosyl DNA phosphodiesterase that repairs topoisomerase- mediated DNA damage. Nature 461, 674–8 (2009). 240. Goodarzi, A. A. & Jeggo, P. A. The repair and signaling responses to DNA double- strand breaks. Adv. Genet. 82, 1–45 (2013). 241. Chapman, J. R., Taylor, M. R. G. M. & Boulton, S. J. S. Playing the end game: DNA double-strand break repair pathway choice. Mol. Cell 47, 497–510 (2012). 242. Pan, M.-R. et al. Chromodomain helicase DNA-binding protein 4 (CHD4) regulates homologous recombination DNA repair, and its deficiency sensitizes cells to poly(ADP-ribose) polymerase (PARP) inhibitor treatment. J. Biol. Chem. 287, 6764–72 (2012). 243. Miller, K. M. et al. Human HDAC1 and HDAC2 function in the DNA-damage response to promote DNA nonhomologous end-joining. Nat. Struct. Mol. Biol. 17, 1144–51 (2010). 244. Smeenk, G. et al. The NuRD chromatin-remodeling complex regulates signaling and repair of DNA damage. J. Cell Biol. 190, 741–9 (2010). 245. Bagchi, A. et al. CHD5 is a tumor suppressor at human 1p36. Cell 128, 459–75 (2007). 246. Bagchi, A. & Mills, A. a. The quest for the 1p36 tumor suppressor. Cancer Res. 68, 2551–6 (2008). 247. Wu, X. et al. Chromodomain helicase DNA binding protein 5 plays a tumor suppressor role in human breast cancer. Breast Cancer Res. 14, R73 (2012). 248. Wang, L. et al. Downregulation of chromatin remodeling factor CHD5 is associated with a poor prognosis in human glioma. J. Clin. Neurosci. 20, 958–63 (2013). 249. Wang, X., Lau, K. K. K., So, L. K. Y. & Lam, Y. W. CHD5 is down-regulated through promoter hypermethylation in gastric cancer. J. Biomed. Sci. 16, 95 (2009). 250. Wang, J. et al. The involvement of CHD5 hypermethylation in laryngeal squamous cell carcinoma. Oral Oncol. 47, 601–8 (2011).
164 251. Van Maerken, T. et al. Functional analysis of the p53 pathway in neuroblastoma cells using the small-molecule MDM2 antagonist nutlin-3. Mol. Cancer Ther. 10, 983–93 (2011). 252. Geisbrecht, B. V & Gould, S. J. The human PICD gene encodes a cytoplasmic and peroxisomal NADP(+)-dependent isocitrate dehydrogenase. J. Biol. Chem. 274, 30527–33 (1999). 253. Chou, D. M. et al. A chromatin localization screen reveals poly (ADP ribose)- regulated recruitment of the repressive polycomb and NuRD complexes to sites of DNA damage. Proc. Natl. Acad. Sci. U. S. A. 107, 18475–80 (2010).
165
Appendix A - Defining the Relocalization of CHD5 to DNA Damage Sites
9.1 Preface
This appendix contains a summary of early thesis work that was focused on the chromatin remodelling protein CHD5. This work was undertaken due to an interest in the chromatin context of DNA damage signaling and repair, with focus on how chromatin remodelers influence the outcome of IR-induced DNA damage detection, signaling and repair. Unfortunately, due to technical limitations and numerous setbacks in progress, I made the decision to largely abandon this work. A major factor in this was the discovery that CHD5 is essentially deleted or not expressed in nearly all tractable mammalian cell culture lines. The decision to stop pursuing the analysis of CHD5 was taken following a serious consideration of whether the work could be consolidated in a timely manner. A summary of this work is included here only as an appendix, as the indeed this data not fully consolidated at this time and does not narratively fit in the overall work presented in earlier chapters.
166 9.2 Introduction
9.2.1 DNA double strand break repair and chromatin
To understand how chromatin modifiers such as CHD5 tie into DNA repair, it is important to recall the current models for DSB repair, the form of damage in which the influence of the specific chromatin context has perhaps been best studied. Rather than reiterating the details of this yet again, I refer the reader to my introductory chapter in this thesis for a summary of human DSB repair pathways and processes. For an even more comprehensive review of all these pathways, please see (59,240). I have also authored an extensive review about chromatin and nucleosome dynamics, and so refer the reader to that article for a comprehensive review of that subject (Stanley et al. 2013)63. Chromatin context is relevant for DSB repair and, in a general sense, enzymatic chromatin remodeling activity is important for certain key steps, including DSB end processing and resection, but is also specific to certain chromatin environments (e.g. heterochromatin) that require ‘special treatment’ to enable efficient repair101,241. To support the idea that chromatin context can complicate repair processes or may require specialized pathways, there is a higher somatic mutation rate in cancer cells at loci with epigenetic markers for heterochromatin (eg.H3K9me3-rich) 110. Current evidence also suggests that the DSB end processing phase of NHEJ, as well as the DNA end resection and Rad51 filament formation stages of HR, all require a degree of nucleosome remodeling and chromatin relaxation99,100,241. To illustrate the relevance of chromatin regulation to DSB repair, let us look at CHD3 and CHD4 as examples. These are the two most similar proteins to CHD5 (in terms of percent sequence homology) and each has an established role in the DDR (DNA Damage Response). Isoform 1 of CHD3 (CHD3.1) is bound to the KAP-1 transcriptional co-repressor protein, which tethers it to the chromatin and localizes its activity to enable nucleosome compaction and heterochromatin formation. During the DSB response, ATM-mediated phosphorylation of KAP-1 leads to CHD3.1 dispersal from heterochromatic nucleosomes. This enables chromatin relaxation, facilitates γH2AX (H2AXS139p, a DSB-induced histone modification) signal expansion and is required for DSB repair in regions of KAP-1 rich heterochromatin138. CHD4 interacts with MCPH1, a protein regulating recruitment of
167 RAD51 and BRCA2 to damage sites242. CHD4 deficiency confers PARP inhibitor sensitivity, strongly suggesting CHD4 as an import regulator of HR242. CHD4 is also important for the recruitment of the histone deacetylases HDCA1/2 to DSBs, where they promote NHEJ via de-acetylating H3K56ac marks243. It is also thought that CHD4 chromatin remodeling is important to RNF8 mediated recruitment and subsequent BRCA1 and RNF168 recruitment events71. CHD4 is notable as being the first CHD-class chromatin remodeling enzyme to be found to dynamically relocalize to DSBs induced by laser microirradiation244.
9.2.2 CHD5
CHD5 is part of the Chromodomain Helicase DNA-binding (CHD) family of ATP- dependent chromatin remodeling enzymes (diagram below), proteins that can alter the spacing between nucleosomes by using the energy of ATP to adjust DNA and histone linkages. Unlike any other CHD enzyme (which are expressed ubiquitously), CHD5 is most highly expressed in brain tissue, suggesting a role in neurological development and/or function245,246. CHD5 is considered a class II CHD enzyme, by its similarity to CHD3 and CHD4. These three enzymes are all (mutually exclusive) components of Nucleosome Remodeling and Deacetylase (NuRD) complexes, multi-subunit complexes
Figure A-1 Diagram of the amino acid sequence of CHD5 highlighting main protein domains that regulate chromatin formation and transcription. Whether CHD5 contributes to genomic stability is not known, although this is considered possible, even likely, given its established linkage to the p53 pathway245. At this point, however, it is uncertain whether the p53 dysfunction observed in CHD5 defective cells is a direct or indirect consequence of CHD5 loss (for example, p53 inactivation is a common result of genomic instability) (29). A 1p36 chromosomal
168 deletion occurs in 30-35% of neuroblastoma (NB) cases (a solid neurological tumor most prevalent in children), and this correlates with advancing (and more difficult to treat) disease and is an emerging prognostic marker for disease outcome (for neuroblastoma as well as several other cancers). A significant step forward in our understanding of neuroblastoma was taken in 2007, when Bagchi et al. identified the critical tumor suppressor gene deleted from 1p36 in neuroblastoma: the ATP-dependent chromatin remodeling enzyme CHD5245,246. Since then, loss of CHD5 expression has since been identified in a wider array of cancer types, with epigenetic silencing of the CHD5 gene shown to be an important part of CHD5 protein ablation during cancer development (1p36 deletions are commonly observed only at one allele, with DNA methylation and silencing of the remaining CHD5 allele leading to total loss of CHD5 expression)247–250. Since 2007, however, almost no new information has emerged on how CHD5 functions as a tumor suppressor. What is known has come from the original analysis of the 1p36 deletion of CHD5 and CHD5 mutant mice245. As expected, based on the human disease, CHD5 deficiency proved to be highly transforming in MEF (Mouse Embryonic Fibroblasts) model systems and also promoted murine tumorigenesis in vivo. While total CHD5 deletion was embryonic lethal, CHD5 haploinsufficiency lead to rapid tumor formation in multiple organs and was observed to correlate with a lack of normal cellular apoptosis and/or senescence. Fitting with a role for CHD5 in pathways controlling apoptosis or senescence, mice over-expressing CHD5 displayed an absence of tumors, with an early aging phenotype containing abnormally high levels of premature cell death or permanent cell cycle arrest/exit. These phenomena were linked to p53 pathway alterations as CHD5 levels were manipulated, possibly via deregulation of p16/ARF, although precisely how CHD5 loss or gain leads to this was not determined. In 2009, Maerken et al. investigated the function of p53 in NB and found that NB often demonstrates a relatively intact p53 pathway, meaning the genes encoding p53 or MDM2 are rarely mutated251. Yet these tumor cells need to somehow inactivate the p53 pathway to avoid the failsafe checkpoint mechanisms it controls252. They suggest several possible mechanisms for non-mutational p53 inactivation, one of which is the suppression of the p14ARF-p53 signaling axis by loss of CHD5. Their hypothesis emerged from a separate study that showed that loss of CHD5
169 via shRNA silencing compromised p53 and promoted tumorigenesis. Ultimately, this provides evidence to the need for CHD remodeling activity at the Cdkn2a locus (p16 + p19) to activate transcription of these genes. p14ARF can then inhibit MDM2 and by effect release p53 from MDM2 suppression, or to state this more simply loss of CHD5 leads to p53 suppression246. Senescence and apoptosis are important for preventing cancer and are also strongly affected by the intrinsic genome stability of a cell 244. Given that the other class II CHD enzymes, CHD3 and CHD4, are known to impact genomic stability by regulating the DSB response, this supported the notion that CHD5 may also and that this could underlie its function (i.e. genomic instability increases in the absence of CHD5, leading to mutations that cause apoptosis or senescence failure and neoplastic transformation). This possibility is supported by CHD5’s known status as a tumor suppressor and the high degree of sequence and predicted functional similarity of CHD5 to CHD3 and CHD4. The similarity between CHD5 and CHD4 is notable, as CHD4 is dynamically recruited to laser microirradiation-induced DSB tracks, has been shown to play a role in DSB repair processes and is itself modified post-translationally during the DSB response. At the time I commenced my graduate studies, there was no indication whether CHD5 was dynamically altered during the DSB response.
9.3 Results
9.3.1 Endogenous CHD5 expression is difficult to demonstrate and antibody specificity could not be demonstrated.
There were no commercially available antibodies that were proven to be selective and specific to CHD5 at the outset of my thesis work. To develop such a tool, I worked in conjunction with a company, ABPro, to produce CHD5 specific antibodies. ABpro use proprietary design strategies to choose two peptides to raise the antibodies (regions show on diagram above): • Peptide 1: CQRPVTPIPDVQSSKGGNLAASAKKKH • Peptide 2: EGQSGRRQSRRQLKSC
170 CHD5 is 1954 amino acids long, and peptide 1 spans 1213-1238, while peptide 2 spans 1388-1402. The 1206-1253 region, which contains peptide 1, is predicted to be an unstructured region. The region spanning amino acids 1377-1531 is attributed to the conserved DUF1086, a domain of unknown function seen in the CHD family of proteins (figA-1). The company inoculated mice (one animal per peptide) with these peptides eight times and performed initial test bleeds and titre assays. Isolated B cells were then fused to myeloma cells. Another screen was then carried out on the 768 isolates from each of the mice. The supernatants from the top 72 were sent to use for testing, and I tested these using cell extracts from HeLa cells with ectopic GFP-tagged CHD5 by WB and also in IF. I indicated to ABPro our top five choices, they then I subcloned these cells, performed a final ELISA screen, and the top 11 subclones were then sent to us, along with a supernatant sample from each culture. I grew the hybridoma and purified antibody from the cell media using the relevant peptide conjugated to NHS sepharose beads and eluted using glycine. Purified antibodies were tested against cell extracts from HEK293 cells with ectopic GFP-tagged CHD5 and IP purified CHD5-GFP from the same extracts. 7A10-E9 anti-CHD5 antibody provided the best result, yet it did not cross react with endogenous CHD5 in any tested cells when used at 1:20 dilution against 50 µg of RIPA prepared whole cell extracts (or mouse brain cortex extract) (figA-2a). However, when the glycine elution of concentrated antibody was tested undiluted against a 5e7 cells lysed directly in Laemelli buffer, the 7A10-E9 anti-CHD5 antibody did detect a signal at the expected molecular weight. The undiluted antibody also detected a band at the expected size in extracts from HEK293 overexpressing CHD5-GFP, and siRNA knockdown using CHD5-targeting siRNA did suppress the signal. However, when the undiluted preparation was used in HEK293 the signal was also lost with siRNA targeted against CHD3 and CHD4. Notably, CHD4 has a nearly identical expected molecular weight to CHD5, and I was forced to conclude that the antibody may also be detecting CHD4. Ultimately, the antibody work was not pursued as the use on undiluted 7A10-E9 anti-CHD5 antibody required unfeasibly large volumes of preparation that degraded upon freeze thaw and displayed considerable batch variation.
171 Figure A-2 – (a) a panel of cell lines and tissue was probed using the custom CHd5 antibody, the only clear band at 250kDA is in the overexpression sample. (b) large amounts of protein were loading from samples directly lysed into sample buffer and again probed with the CHD5 Ab (undiluted elute) possible detect of endogenous levels inf HeLa and U2OS cells.
In summary, it was “more trouble that it was worth” when weighed against more profitable scientific avenues of investigation.
9.3.2 CHD5 knockdown did not impact DSB repair kinetics in human cell lines
It was only possible to validate the siRNA targeted at CHD5 using cells with CHD5 overexpression, but (at least under those conditions) it appeared to be effective (figA-3). Using this knockdown strategy in human fibroblasts, I carried out a γH2AX foci enumeration analysis to explore DSB induction and repair kinetics in the absence of CHD5. In 1BR-hTERT cells transfected with siCHD5 or siRNF8 (a positive control for a DSB repair defect), irradiated with 2 Gy of gamma-radiation. There was no observable impact in the absence of CHD5 (fig A-4). However, this was not conclusive as the 7A10- E9 anti-CHD5 antibody was not able to detect a baseline signal for CHD5 expected size in these cells (fig A-2). Indeed, CHD5 may not have been expressed at all in this line, and thus the experiment was not conclusive. DSB repair could not be monitored in the highly
172 transformed HEK293 cell line that I was able to detect endogenous CHD5, as they contain a vast amount of endogenous damage, and generally perturbed for many DDR pathways and are considered “useless” within DSB repair scientific circles for such an analysis.
9.3.3 CHD5 does recruit to laser microirradiation tracks, but no dependency for this process were found Figure A-3 – (a) Cells transfected with GFP-CHD5 and/or siCHD5 were probed with the CHD5 Ab. Overexpressed protein is detectable at the expectd size, this band is lost with siCHD5 and is reduced with I also examined the the control siKAP1 (b) Cells transfected with GFP-CHD5 or dynamics of CHD5 using siCHD4/5/6 were probed with the CHD5 Ab, the endogenously detected band(lane1) is again lost with siCHD5, however is also lost laser micro-irradiation with siCHD4 and siCHD3 treatments. induced DSB recruitment live cell imaging assays. A calibration of the assay was carried out in HeLa cells over- expressing CHD5GFP (fig A-5), and it was clear that CHD5 was rapidly and convincingly recruited to laser tracks. I examined a series of small molecule inhibitors and siRNAs (targeting known DDR processes and enzymes in DSB repair) in order to determine the molecular context of CHD5 recruitment to sites of microirradiation induced DNA damage. However, I failed to find any condition upon which CHD5 recruitment was dependent (fig A-6). It is possible that CHD5 is directly recruited to DSBs, however this remains to be consolidated. It was notable that PARP inhibition did NOT preclude CHD5
173 recruitment as all other CHD enzymes that are known to exhibit this behavior have been shown to be PARP-sensitive253.
Figure A-4 – (a)IF image of 48BR primary human fibroblasts irradiated with 3Gy, fixed 30min post IR and stained for γH2AX and 53BP1 showing the gamma radiation induced DNA damage foci which can be enumerated to measure the kinetics of DNA repair. (b) enumerating IRIF in 48BR primary human fibroblasts treated with Mock siRNA (siMock) as a negative control, CHD5 siRNA (siCHD5) or RNF8 siRNA (siRNF8) as a positive control, as RNF8 knockdown has a known repair defect.
174
(a)
19 (b) GFP-CHD5 Recruitment Quantification (Zen) 17
15
13
11
9
7 Realtive Change in Pixel Intensity 5
3
1 0 1 2 3 4 5 6 7 8 9 1011121314151617181920 Time After Irradiation (min)
Figure A-5 – (a) Live cell images of HeLa showing CHD5-GFP laser microirradiation recruitment.(b) Quantification of CHD5-GFP laser microirradiation recruitment. n=30, err=SD
175
Figure A-6 – (a) Live cell images of HeLa showing CHD5-GFP laser microirradiation recruitment after treatment with a panel of inhibitors or siRNAs
176 CHD5 Nucleosome and BAF Interactions I next examined potential interactions of CHD5 with chromatin and other chromatin regulatory factors. By overexpressing CHD5GFP in HEK293 cells and immunoprecipitating CHD5 with and anti-GFP antibody, I examined whether specific chromatin marks and certain epigenetically modified histones were present in the IP CHD5 extracts. Cells were treated ± 8 Gy gamma IR 1 hr before cell lysis, to examine the influence of the DDR on interactions. In the CHD5 IP, there was a high background of H2AX signal even in the unirradiated samples; in fact, in the first attempt, there was no increase in signal after IR at all – a common problem of working with HEK293 cells
Figure A-7 – IP from cells transfected with GFP-CHD5 and/or GFP and treated with IR (gamma 10Gy). Pulldowns with anti-GFP antibodies were probed for multiple histone tail PTMs to test for IR responsive chromatin interactions.
177 which, as mentioned above, have a very high level of endogenous DSBs that swamps any additional signal caused by exogenous DNA damage. In a repeat of this experiment (fig A-7), I observed an increased interaction between CHD5GFP and a number of specific histone epigenetic marks (H3K9me3, H4K8ac) after IR. I also reproduced the previously reported HDAC1 interaction with CHD5, however in this case I saw no difference in the protein pulldown levels after IR. In an attempt to resolve this, I repeated a small-scale experiment with RIPA lysis in an effort to reduce background and increase stringency. Although no conditions were altered, there was no signal in the no-IR GFP group. CHD5GFP Figure A-8 – (a) Input of IP; Cells transfected with GFP-CHD5 and/or the overexpression alone Flag tagged components of the BAF complex were probed with the Flag Ab to confirm expression (b) Pulldowns with anti-GFP antibodies were probed appeared to simulate for Flag to test for a CHD5-BAF interaction. γH2AX signal.
178 Another line of inquiry followed hits from a mass spectrometry analysis of IP CHD5GFP that was aimed at identifying novel proteins interacting with CHD5. That analysis was carried out in HEK293 cells, using GFP antibodies to IP CHD5GFP and was performed by my colleague Dustin Pearson in the lab. The hits identified by mass spectrometry in the IP CHD5 were filtered through the so-called “crapome” to eliminate off-target hits, and then revealed several components of the BAF chromatin remodelling complex being present in the CHD5 immunoprecipitates. To pursue these interactions, I co-expressed several FLAG-tagged components of BAF complex with GFP tagged CHD5 in HEK293 cells and carried out reciprocal FLAG and GFP pulldowns an immunoblot to determine whether these interactions could be observed. In short, I was unable to reproduce the interaction between any BAF component and CHD5 after multiple attempts (fig A-8). One interesting result from these attempts was apparent co stabilization of CHD5 and SMARCA4 when they were co- expressed (fig A-9), however the significance of this remains unclear.
Figure A-9 – (a) Cells transfected with GFP-CHD5 and/or Flag-SMARCA4 were probed with GFP Ab. Overexpressed protein is detectable at the expected size, the CHD5 band is stronger with SMARCA4 co-expression (b) Cells transfected with GFP-CHD5 and/or Flag-SMARCA4 were probed with Flag Ab. Overexpressed protein is detectable at the expected size, the SMARCA4 band appears stronger with CHD5 co-expression
179 9.4 Conclusions
The antibody work generated a potentially useful reagent but failed to give reproducible results due to numerous technical limitations. This may be due to the difficulty in producing an antibody that will specifically detect CHD5 and not CHD4 – indeed, I found this to be a consistent problem with commercially available CHD5 antibodies early on in my PhD work. The difficulty in antibody reproducibility might benefit from a future troubleshooting of the solution and temperature in which the antibody elutions are prepared. I also suggest that a much larger scale of hybridoma preparation might improve outcomes. Based on Coomassie stains (not shown) the elutions were clean preparations of antibody; however, contaminants may be an issue, and despite considerable efforts to optimize this, it is possible that some further refinements for the immunoblot protocol might increase the usability of the antibody. It is somewhat difficult to interpret these results to make any strong statement on the levels of CHD5 expression in the cell lines that I tested due to the nuances of the antibody issues. However, I found little to no expression in any of the tested lines. Since it is thought that CHD5 is principally expressed in the central and peripheral nervous system, this is perhaps not surprising. The lack of CHD5 signal in the mouse brain tissue sample may be explained by species specificity issues, although at the early stage of peptide selection I were mindful to select regions that were conserved between human to mice as best possible. One could take an mRNA approach to screen for cells that might express CHD5 at higher levels, however that doesn’t necessarily translate to protein levels. The uncertainty in normal expression status of CHD5 precluded interpretation of the γH2AX foci assay, which did not reveal any DSB repair defect after CHD5 knockdown – but this might be because there was no CHD5 being expressed in the first place or could also simply be a negative result. CHD5 tumor suppressive action appears to be through gene expression pathways in the mouse model, and the biochemical mechanism of its chromatin remodelling does differ from other enzymes of the family, so it is plausible that it has no role to play in the repair of DNA double strand breaks. Overexpressed CHD5 was convincingly recruited to laser microirradiation tracks. However, based on the results of these studies, I am unable to conclude whether CHD5 recruitment is an artefactual result or not. It is possible that over-expressing a chromatin
180 remodeler (to the extent endemic to ectopic cDNA expression produces) that is typically expressed at very low levels in a normal cell could disrupts protein complex compositions. CHD5 might be thus be over-represented in multi-protein complexes or even interact with complexes it typically would not. For example, CHD5 is thought to occur in complex with the HDAC1 protein, which is also recruited to laser microirradiation tracks. While CHD5 does appear to interact with nucleosomes in an IR responsive manner, this experiment also relied on heavy overexpression of the protein. This is still an interesting result worth pursuing, attempting a reverse IP of ChIP experiment could be a good way explore these preliminary results further. Additionally, the co-stabilization of CHD5 with BAF complex components is an interesting result. I was unable to establish a convincing interaction between these two proteins, so by what mechanism they might co stabilize remains unclear.
181 Appendix B – Cell Cycle Staging Based on DAPI
Cell Cylce Histogram Derived from Dapi Intesntiy Measures 70
60
50
40
30 Frequency
20
10
0 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 Bucketed Integrated Dapi Values
Appendix B Use of integrated DAPI signal(x-axis) and frequency (y-axis) to chart the cell cycle graph of an asynchronous population of A549 cells grown in the glass bottom 96wp and imaged as per the alpha irradiation protocol.
182
Appendix C - Alpha and Gamma Irradiated A549 Cells
Appendix C Alpha irradiated A549 cells (red: 0.8 Gy), gamma irradiated cells (blue 1.8 Gy ), were allowed to recover for the designated time periods. Cells were fixed, stained with an antibody to detect ɣ-H2AX, and ɣ-H2AX was automatically imaged and analyzed by individual cell. Error bars represent standard error of the mean.
183 Appendix D – Chronic Low Dose Alpha Irradiation of Yeast Cells
Appendix D Alpha irradiated WT yeast cells (top) and RAD50 deletion mutant yeast cells exposed to the alpha sources for the designated period. NOTE: Americium source dose rate was estimated at 0.4gy/min ( yeast culture contributed by J Cobb)
184 Appendix E - Real Time Radon Data
Real time radon reading fluctuates and may be affected by weather, below is the real time radon data collected from a single house in Calgary, with the radon reading and out door temperature both charted.
185 Appendix F – Geospatially Divided Radon Results
The following three tables detail the radon testing results as well as post hoc analysis divided up by the ED and FSC geospatial divisions
Table F-1 - Radon levels by Electoral Districts Metrics OBSERVED RADON CONNECTRATION (Bq/m3)
Electoral District n Geometric Arithmetic Min - Max 99 % CI Mean Mean Alberta 9507
Banff--Airdrie 622 146 110 15-2116 (132,160) Battle River--Crowfoot 65 143 125 21-323 (123,163) Bow River 187 120 100 18-386 (108,132) Calgary Centre 545 89 70 15-962 (82,96) Calgary Confederation 1016 118 82 15-3227 (105,131) Calgary Forest Lawn 126 116 90 15-689 (97,135) Calgary Heritage 700 121 102 15-1040 (114,128) Calgary Midnapore 716 125 104 17-1014 (117,133) Calgary Nose Hill 585 146 107 15-1575 (130,162) Calgary Rocky Ridge 948 125 97 15-1513 (116,134) Calgary Shepard 615 118 99 15-1225 (110,126) Calgary Signal Hill 978 154 120 15-1562 (144,164) Calgary Skyview 137 114 97 15-350 (101,127) Edmonton Centre 131 148 114 15-1151 (120,176) Edmonton Griesbach 38 143 126 39-311 (116,170) Edmonton Manning 44 84 68 20-344 (60,108) Edmonton Mill Woods 44 125 95 26-576 (84,166) Edmonton Riverbend 247 116 97 22-551 (105,127) Edmonton Strathcona 199 152 125 15-582 (136,168) Edmonton West 93 97 81 15-374 (82,112) Edmonton--Wetaskiwin 90 102 87 15-365 (87,117) Foothills 483 177 124 15-7199 (139,215) Fort McMurray--Cold Lake 27 95 81 16-269 (68,122)
186 Grande Prairie--Mackenzie 38 91 71 23-576 (56,126) Lakeland 38 134 107 37-479 (96,172) Lethbridge 102 114 97 15-308 (99,129) Medicine Hat--Cardston-- 82 111 89 17-501 (90,132) Warner Peace River--Westlock 26 141 114 22-418 (96,186) Red Deer--Mountain View 106 140 112 15-617 (117,163) Red Deer--Lacombe 73 145 114 15-658 (117,173) St. Albert--Edmonton 113 134 113 15-394 (117,151) Sherwood Park--Fort 178 121 101 26-651 (105,137) Saskatchewan Sturgeon River--Parkland 60 120 101 15-374 (99,141) Yellowhead 55 142 112 24-411 (110,174)
187 Table F-2 - Radon levels by Province
Metrics OBSERVED RADON CONNECTRATION (Bq/m3)
Geometric Arithmetic Min - Max 99 % CI Electoral District n Mean Mean SK 9507
Battlefords--Lloydminster 55 226 169 30-1271 (156,296)
Cypress Hills--Grasslands 323 290 205 15-2985 (245,335) Desnethé--Missinippi-- 14 138 110 20-350 (83,193) Churchill River Carlton Trail--Eagle Creek 100 166 117 15-1394 (123,209) Moose Jaw--Lake Centre-- 70 181 138 15-774 (142,220) Lanigan Prince Albert 61 116 84 15-840 (78,154)
Regina--Lewvan 262 309 214 15-2107 (267,351)
Regina--Qu'Appelle 121 239 169 20-1430 (190,288)
Regina--Wascana 229 282 191 15-1930 (234,330)
Saskatoon--Grasswood 211 130 111 15-440 (118,142)
Saskatoon--University 220 136 111 15-820 (120,152)
Saskatoon West 107 130 109 15-400 (114,146)
Souris--Moose Mountain 49 198 155 20-860 (144,252)
Yorkton--Melville 52 167 128 15-640 (126,208)
BC
5 ED 12 329 137 18-1599 (25,633)
NT
1 ED 5 39 29 15-100 (1,77)
ON
2 ED 2 241 123 34-448 (-241,723)
MB
1 ED 2 150 149 140-159 (128,172)
188 Table F-3 – Radon District - AOV post Level by Electoral hoc tests
Pair H-B adj. P Pair H-B adj. P Pair H-B adj. P 47006-47002 0 48001-47007 0 47011-47009 0 47010-47002 0 48002-47007 0 47012-47009 0 47011-47002 0 48003-47007 0 48001-47009 0 47012-47002 0 48004-47007 0 48003-47009 0 48001-47002 0 48005-47007 0 48004-47009 0 48003-47002 0 48006-47007 0 48005-47009 0 48004-47002 0 48007-47007 0 48006-47009 0 48005-47002 0 48008-47007 0 48007-47009 0 48006-47002 0 48009-47007 0 48008-47009 0 48007-47002 0 48010-47007 0 48009-47009 0 48008-47002 0 48011-47007 0 48010-47009 0 48009-47002 0 48012-47007 0 48011-47009 0 48010-47002 0 48013-47007 0 48012-47009 0 48011-47002 0 48014-47007 0 48013-47009 0 48012-47002 0 48016-47007 0 48014-47009 0 48013-47002 0 48018-47007 0 48016-47009 0 48014-47002 0 48019-47007 0 48018-47009 0 48016-47002 0 48020-47007 0 48019-47009 0 48018-47002 0 48021-47007 0 48020-47009 0 48019-47002 0 48022-47007 0 48021-47009 0 48020-47002 0 48024-47007 0 48022-47009 0 48021-47002 0 48026-47007 0 48026-47009 0 48022-47002 0 48027-47007 0 48027-47009 0 48026-47002 0 48029-47007 0 48029-47009 0 48027-47002 0 48030-47007 0 48031-47009 0 48029-47002 0 48031-47007 0 48032-47009 0 48031-47002 0 48032-47007 0 48012-48004 0 48032-47002 0 48033-47007 0 48022-48004 0 48033-47002 0 48004-47008 0 48024-47002 4.4E-07 47007-47004 0 48005-47008 0 59015-48004 1.19E-06 47007-47006 0 48007-47008 0 48008-47008 1.3E-06 47010-47007 0 48010-47008 0 47009-47006 2.1E-06 47011-47007 0 48011-47008 0 48017-47007 6.26E-06 47012-47007 0 47010-47009 0 48034-47007 6.41E-06
189 Pair H-B adj. P Pair H-B adj. P Pair H-B adj. P 48030-47002 1.11E-05 59015-48003 7.66E-05 48025-47007 0.001068 48033-47009 1.46E-05 59015-48008 8.29E-05 59015-48012 0.001078 59015-48020 1.65E-05 59015-48032 9.3E-05 59015-48014 0.001243 59015-48016 1.85E-05 48023-47007 0.00013 59015-48019 0.001331 48018-47008 2.16E-05 59015-47006 0.00018 59015-48030 0.001635 59015-48021 2.75E-05 59015-47010 0.000184 59015-48002 0.001644 59015-48005 3.8E-05 59015-48023 0.000199 48013-47008 0.001662 59015-48011 4.17E-05 59015-48033 0.000261 59015-48034 0.001878 48024-47009 4.29E-05 59015-47012 0.000286 59015-48025 0.001993 47004-47002 4.62E-05 59015-47011 0.000309 48032-47008 0.001997 59015-48018 4.7E-05 59015-48031 0.000401 48002-47009 0.002904 59015-48024 5.13E-05 59015-48001 0.000549 48021-47008 0.00322 59015-48007 5.6E-05 59015-48009 0.000585 48022-48011 0.003491 59015-48013 5.6E-05 48020-47008 0.000601 48004-48001 0.004003 48022-48005 6.16E-05 48017-47002 0.000604 48023-47002 0.00404 48002-47002 6.49E-05 59015-48029 0.000696 59015-48015 0.004189 59015-48026 7.15E-05 59015-48017 0.000715 48009-48004 0.004625 59015-48027 7.28E-05 48003-47008 0.000757 48004-47001 0.004752 59015-48006 7.34E-05 48030-47009 0.000849 59015-48010 7.61E-05 48034-47002 0.000874
190 Table F-4 Radon Level By FSC T3K-S0L 0 T3M-S0N 0 - AOV post hoc tests T3L-S0L 0 T3Z-S0N 0 S0L-S0K 0 T3M-S0L 0 T4B-S0N 0 S4S-S0K 0 T3Z-S0L 0 T5T-S0N 0 S6V-S0L 0 T4B-S0L 0 T6J-S0N 0 S7H-S0L 0 T4C-S0L 0 T6R-S0N 0 S7J-S0L 0 T5T-S0L 0 T8A-S0N 0 S7K-S0L 0 T6J-S0L 0 T8N-S0N 0 S7M-S0L 0 T6R-S0L 0 T2E-S4N 0 S7N-S0L 0 T6W-S0L 0 T2L-S4N 0 T0J-S0L 0 T8A-S0L 0 T2M-S4N 0 T0L-S0L 0 T8N-S0L 0 T2S-S4N 0 T1B-S0L 0 S7J-S0N 0 T2T-S4N 0 T1K-S0L 0 S7K-S0N 0 T2V-S4N 0 T1W-S0L 0 T1K-S0N 0 T2Z-S4N 0 T1X-S0L 0 T1W-S0N 0 T3G-S4N 0 T1Y-S0L 0 T2C-S0N 0 T3L-S4N 0 T2C-S0L 0 T2E-S0N 0 S7J-S4R 0 T2E-S0L 0 T2J-S0N 0 S7K-S4R 0 T2J-S0L 0 T2L-S0N 0 T2C-S4R 0 T2K-S0L 0 T2M-S0N 0 T2E-S4R 0 T2L-S0L 0 T2N-S0N 0 T2J-S4R 0 T2M-S0L 0 T2S-S0N 0 T2L-S4R 0 T2N-S0L 0 T2T-S0N 0 T2M-S4R 0 T2S-S0L 0 T2V-S0N 0 T2N-S4R 0 T2T-S0L 0 T2W-S0N 0 T2S-S4R 0 T2V-S0L 0 T2X-S0N 0 T2T-S4R 0 T2W-S0L 0 T2Y-S0N 0 T2V-S4R 0 T2X-S0L 0 T2Z-S0N 0 T2W-S4R 0 T2Y-S0L 0 T3A-S0N 0 T2X-S4R 0 T2Z-S0L 0 T3B-S0N 0 T2Y-S4R 0 T3A-S0L 0 T3C-S0N 0 T2Z-S4R 0 T3B-S0L 0 T3E-S0N 0 T3A-S4R 0 T3C-S0L 0 T3G-S0N 0 T3B-S4R 0 T3E-S0L 0 T3H-S0N 0 T3C-S4R 0 T3G-S0L 0 T3K-S0N 0 T3E-S4R 0 T3H-S0L 0 T3L-S0N 0 T3G-S4R 0
191 T3H-S4R 0 T2T-S4S 0 T2C-S4T 0 T3K-S4R 0 T2V-S4S 0 T2E-S4T 0 T3L-S4R 0 T2W-S4S 0 T2J-S4T 0 T3M-S4R 0 T2X-S4S 0 T2L-S4T 0 T3Z-S4R 0 T2Y-S4S 0 T2M-S4T 0 T4B-S4R 0 T2Z-S4S 0 T2N-S4T 0 T5T-S4R 0 T3A-S4S 0 T2S-S4T 0 T6R-S4R 0 T3B-S4S 0 T2T-S4T 0 T8A-S4R 0 T3C-S4S 0 T2V-S4T 0 S6V-S4S 0 T3E-S4S 0 T2W-S4T 0 S7H-S4S 0 T3G-S4S 0 T2X-S4T 0 S7J-S4S 0 T3H-S4S 0 T2Y-S4T 0 S7K-S4S 0 T3J-S4S 0 T2Z-S4T 0 S7L-S4S 0 T3K-S4S 0 T3A-S4T 0 S7M-S4S 0 T3L-S4S 0 T3B-S4T 0 S7N-S4S 0 T3M-S4S 0 T3C-S4T 0 S7W-S4S 0 T3R-S4S 0 T3E-S4T 0 T0J-S4S 0 T3Z-S4S 0 T3G-S4T 0 T0L-S4S 0 T4A-S4S 0 T3H-S4T 0 T0M-S4S 0 T4B-S4S 0 T3K-S4T 0 T1B-S4S 0 T4C-S4S 0 T3L-S4T 0 T1K-S4S 0 T4R-S4S 0 T3M-S4T 0 T1P-S4S 0 T5R-S4S 0 T3Z-S4T 0 T1S-S4S 0 T5T-S4S 0 T4B-S4T 0 T1W-S4S 0 T6C-S4S 0 T5T-S4T 0 T1X-S4S 0 T6J-S4S 0 T6J-S4T 0 T1Y-S4S 0 T6M-S4S 0 T6R-S4T 0 T2A-S4S 0 T6R-S4S 0 T8A-S4T 0 T2C-S4S 0 T6W-S4S 0 T2C-S9H 0 T2E-S4S 0 T8A-S4S 0 T2E-S9H 0 T2G-S4S 0 T8H-S4S 0 T2J-S9H 0 T2J-S4S 0 T8L-S4S 0 T2L-S9H 0 T2K-S4S 0 T8N-S4S 0 T2M-S9H 0 T2L-S4S 0 S7J-S4T 0 T2N-S9H 0 T2M-S4S 0 S7K-S4T 0 T2S-S9H 0 T2N-S4S 0 T1K-S4T 0 T2T-S9H 0 T2S-S4S 0 T1W-S4T 0 T2V-S9H 0
192 T2W-S9H 0 T0B-S4S 2.21E-05 T2H-S4S 0.000329 T2X-S9H 0 T4P-S4S 2.27E-05 S6V-S4R 0.000354 T2Y-S9H 0 T2A-S0L 2.74E-05 T5Z-S4S 0.000365 T2Z-S9H 0 T3R-S0L 3.05E-05 S4S-S0G 0.000372 T3A-S9H 0 T4C-S0N 3.27E-05 T4X-S4S 0.000387 T3B-S9H 0 T1X-S4R 4.33E-05 T3P-T3E 0.000407 T3C-S9H 0 T2G-S0L 4.46E-05 T1P-S0L 0.000473 T3E-S9H 0 T3P-T2E 4.69E-05 T4C-S4R 0.000476 T3G-S9H 0 T1Y-S0N 5.56E-05 T5R-S0L 0.000501 T3K-S9H 0 T3J-S0L 6.12E-05 T6H-S4S 0.000501 T3L-S9H 0 S7N-S4R 6.5E-05 T4B-S9H 0.000513 T6R-S9H 0 T2W-S4N 6.67E-05 T3G-T1S 0.000521 T2T-T1S 0 T2X-S4N 6.76E-05 V0B-T2L 0.000646 T3P-T2L 0 T1Y-S4T 6.89E-05 T8H-S0L 0.000702 T3P-T2M 0 T6R-S4N 7.54E-05 V0B-T5T 0.000734 T3P-T2S 0 S4V-S4S 7.95E-05 T3K-S4N 0.000773 T3P-T2T 0 T1J-S4S 0.000105 T1V-S4S 0.00081 T3P-T2Z 0 S6V-S4T 0.000107 T6L-S4S 0.000996 T3P-T3G 0 T2C-S4N 0.000111 T1B-S4T 0.001074 T3P-T3L 0 T2J-S4N 0.000133 V0B-T3G 0.001078 T8N-S4T 7.57E-07 T3C-S4N 0.000141 V0B-T3L 0.001202 T1K-S4R 2.67E-06 V0B-T2T 0.000152 T2M-S4V 0.001254 T1X-S0N 3.9E-06 T2K-S4T 0.000164 T2B-S4S 0.001322 T1W-S4R 4.06E-06 T6E-S4S 0.000174 V0B-T2E 0.001425 T6J-S4R 5.23E-06 T1Y-S4R 0.000193 T2N-S4N 0.001464 T2T-S4X 5.35E-06 T4R-S0L 0.000198 T3P-T2W 0.00149 T3E-S4N 7.36E-06 T2M-T1S 0.00021 T5T-S4N 0.001535 T1X-S4T 7.45E-06 S7T-S4S 0.000224 T0L-S0N 0.001537 S7N-S0N 8.45E-06 S6V-S0N 0.000226 T5T-S9H 0.001545 T3M-S9H 1.06E-05 T3H-S9H 0.000227 T6M-S0L 0.001551 T5N-S4S 1.42E-05 V0B-T2M 0.000231 T3P-T2X 0.001571 T2T-S4V 1.46E-05 V0B-T2S 0.000263 T3Z-S9H 0.001601 S7N-S4T 1.78E-05 T2M-S4X 0.000281 S7V-S4S 0.001645 T8N-S4R 1.83E-05 T2K-S4R 0.000282 T6B-S4S 0.001646 T3P-T2V 1.91E-05 T2Y-S4N 0.000283 T0J-S4T 0.001665 T2K-S0N 2.03E-05 T4A-S0L 0.0003 V0B-S6V 0.001746 T0M-S0L 2.07E-05 T4C-S4T 0.000306 T6R-T3P 0.001794
193 V0B-T2C 0.001816 V0B-T2V 0.001878 V0B-T2Z 0.002097 T1B-S0N 0.002214 S7L-S0L 0.002249 T3L-T1S 0.002495 T3P-T2C 0.002561 T8L-S0L 0.002564 T0J-S0N 0.002569 T3P-T2J 0.002581 V0B-T6R 0.002619 V0B-T3C 0.002716 V0B-T1K 0.002748 T3B-S4N 0.002751 S7M-S4T 0.003025 T0L-S4T 0.003113 T3G-S4X 0.003113 T3P-T3C 0.003124 T1B-S4R 0.003274 T5Y-S4S 0.003569 T6W-S4T 0.003674 V0B-T2X 0.004055 V0B-T3E 0.004231 T0J-S4R 0.004765
194
Appendix G – Bill 209
Presented here is a letter from Robyn Luff referring to bill 209, the full bill was accessible as of December 2018 at: http://www.assembly.ab.ca/ISYS/LADDAR_files/docs/bills/bill/legislature_29/session_3/2 0170302_bill-209.pdf
195
196
197
198
199
200