UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Factors influencing the sensitivity of the alkaline comet assay for detection of DNA damage after low… Fjell, Christopher David 1995

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1995-0441.pdf [ 4.95MB ]
JSON: 831-1.0085120.json
JSON-LD: 831-1.0085120-ld.json
RDF/XML (Pretty): 831-1.0085120-rdf.xml
RDF/JSON: 831-1.0085120-rdf.json
Turtle: 831-1.0085120-turtle.txt
N-Triples: 831-1.0085120-rdf-ntriples.txt
Original Record: 831-1.0085120-source.json
Full Text

Full Text

FACTORS INFLUENCING THE SENSITIVITY OF THE ALKALINE COMET ASSAY FOR DETECTION OF DNA DAMAGE AFTER LOW DOSES OF RADIATION by CHRISTOPHER DAVID FJELL B.A.Sc, The University of British Columbia, 1990 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE ' ' in THE FACULTY OF GRADUATE STUDIES Department of Physics We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August 1995 © Christopher David Fjell, 1995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia Vancouver, Canada DE-6 (2/88) 11 Abstract: The alkaline comet assay is a sensitive technique for measuring single-strand breaks in DNA of individual mammalian cells. The method uses fluorescent microscopy to examine DNA patterns resulting from alkaline agarose gel electrophoresis of individual cells after staining with propidium iodide. Factors influencing the sensitivity of the assay to low doses of radiation were investigated in three parts. 1) A model was developed for the digital image supplied by the camera; and improvements in image quality were achieved by subtracting the dark-field image. 2) The conditions for staining of comets with propidium iodide were optimized for image quality by minimizing the background fluorescence through inclusion of 0.02 M NaCl, increasing in the staining time to several hours, and mimmizing the dye concentration. The dependence of comet fluorescence on dye concentration required a model of two binding modes with equilibrium constants, 3.7±0.6 xlO5 M-' and 3.6±0.2 xlO7 M- 1 . These values are in general agreement with reported values for DNA in solution (3.0±0.2 xlO5 M"1) and bound by nucleoprotein (1.1±0.4 xlO7 M"1). A decline in the number of binding sites with increasing radiation damage explained the loss of comet fluorescence with increasing DNA damage. 3) Limitations in sensitivity due to background damage were investigated. A portion of background damage was found to be induced during cell suspension and preparation of agarose gels. The measurement of damage was not significantly influenced by alkaline lysis duration but was strongly affected by extended alkaline rinse duration. Enhanced sensitivity and minimum background damage were found using a cell suspension including culture medium, higher temperature agarose stock solution and a three hour rinse protocol. The sensitivity of the assay was found to be limited by additional background variations at an average level equivalent to 0.05 Gy. Sensitivity below this level would require averaging measurements from multiple, replicate samples. U l T A B L E O F C O N T E N T S Abstract ii Table of Contents iii List of Tables v List of Figures vi Acknowledgements ix 1. Introduction 1.1 The Significance of DNA Damage 1 1.2 Overview of the Alkaline Comet Assay 2 2. General Methods and Materials 2.1 Cell Culture and Handling 6 2.2 Irradiation Conditions and Dosimetry 6 2.3 Comet Assay Technique 11 2.4 Statistical Analysis 13 3. Development of Software for Comet Analysis 3.1 Introduction and Obj ectives 14 3.1.1 Design Obj ectives 14 3.1.2 Model of Camera Response 14 3.2 Methods and Materials 18 3.3 Results and Discussion 18 3.3.1 Experimental Evaluation of Camera Model 18 3.3.2 Definition of Tail End 25 3.4 Definition of Comet Values 32 3.5 Comparison of New Software with Established Software 40 4. Fluorescent Staining and Measurement Conditions 4.1 Introduction 45 4.2 Methods and Materials 50 4.3 Results 54 4.3.1 Comet Stability and Measurement Error 54 4.3.2 Staining Time and Ionic Strength of Solution 55 4.3.3 Stain Concentration 5 8 4.3.4 pH of Staining Solution 70 4.3.5 Comparison of PI to YOYO 72 4.3.6 Camera Settings 73 iv 4.3.7 Effect of Pre-Stain Irradiation 74 4.4 Discussion 82 Appendix 4.1 The Rate of Alkali Denaturation of DNA 84 5. Effects of Assay Technique on Measurement of Induced Damage 5.1 Introduction 85 5.2 Methods and Materials 86 5.3 Results 87 5.3.1 Cell Suspension and Light Exposure 87 5.3.2 Gelling Technique 87 5.3.3 Duration of Alkaline Lysis and Rinse 90 5.3.4 Summary of Apparent Damage Rates from Different Sources 101 5.4 Discussion 102 6. Conclusions 105 Bibliography 107 V LIST OF TABLES Table 2.1 Dosimetry Values Table 3.1 Measured Fluorescence for Thick and Thin Agarose Gel Table 4.1 Comet Stability with Storage Table 4.2 Point Measurement Error for Comets of Different Levels of Damage Table 4.3 DNA-Dye Binding Parameters Table 4.4 Variability of Comet Measurements Due to Staining (Coefficient of Variation) Table 4.5 Comparison of Staining with YOYO and PI Table 5.1 The Effect of Alkaline Rinse Duration on Tail Moment Measurements Table 5.2 Apparent Damage Rates Due to Assay Protocol Timing vi LIST OF FIGURES Figure 1.1 Fluorescent Image of a 1 Gy Comet Figure 2.1 Electrometer Readings for Different Exposure Times Figure 3.1 The Distribution of Fluorescence for a Low Illumination Blank Field of View Figure 3.2 The Distribution of Fluorescence for a High Illumination Blank Field of View Figure 3.3 The Linearity of Camera Response Figure 3.4 The Dark-Field Image for a Camera with Shutter Closed Figure 3.5 The Distribution of Dark-Field Element Intensity with Time Figure 3.6 The Change in Comet Fluorescence with Camera Useage Figure 3.7 Changes in Comet Fluorescence with Order of Measurement Figure 3.8 The Outline of a Slightly-Damaged Comet Image Above the Image Threshold Figure 3.9 The Outline of a 1 Gy Comet Image Above the Image Threshold Figure 3.10 The Effect of Tail End Size and Threshold on Tail Length for a 1 Gy Comet Figure 3.11 The Effect of Tail End Size and Absolute Threshold on Tail Length for a 1 Gy Comet Figure 3.12 The Effect of Tail End Size and Threshold on Tail Length for Low Damage Comet Figure 3.13 Figure 3.12 The Effect of Tail End Size and Absolute Threshold on Tail Length for Low Damage Comet Figure 3.14 Diagram of Comet Measurements Figure 3.15 Comparison of Comet Measurements Using the New and Established Software for High Damage Figure 3.16 Comparison of Comet Measurements Using the New and Established Software for Low Damage Figure 3.17 Comparison of Comet Measurements Using the New and Established Software for an Asynchronous Cell Population Vll Figure 4.1 The Effect of Staining Duration on Background Fluorescence Figure 4.2 The Effect of Staining Duration on Comet Fluorescence Figure 4.3 The Effect of NaCl and EDTA on Comet Staining Figure 4.4 The Effect of PI Concentration on Tail Moment Measurements Figure 4.5 The Effect of PI Concentration on Tail Length Measurements Figure 4.6 The Effect of PI Concentration on Area Measurements Figure 4.7 Comet Fluorescence Measurement Variation with Stain Concentration Figure 4.8 Image Threshold Variation with PI Concentration Figure 4.9 The Effect of PI Concentration on the Difference Between the Average Comet Fluorescence and the Threshold Figure 4.10 The Effect of PI Concentration on the Comet Fluorescence to Threshold Ratio Figure 4.11 Comet Fluorescence for Varying Dye Concentration Figure 4.12 The Effect of pH of Stain on Comet Fluorescence Figure 4.13 The Effect of Gain Setting on the Dim Comet Signal Compared to Threshold Figure 4.14 The Effect of Gain Setting on Dim Comet Measurements Figure 4.15 The Effect of Irradiation Before Staining on Measurements of Asynchronous V79 Cells Bivariate Plot Figure 4.16 The Dose-Response for Slides Irradiated Before Staining Figure 4.17 The Effect of Irradiation of Slides Before Staining on the Signal to Threshold Ratio and Difference Figure 4.18 Modification of the Cell-Cycle Measurement by Irradiation before Staining Figure 5.1 Apparent Damage Due to Cell Suspension Figure 5.2 Apparent Damage With Cooling of Agarose During Mixing Figure 5.3 Apparent Damage for Varying Lysis Duration Figure 5.4 Apparent Damage Due to Rinse Duration Figure 5.5 The Dependence of the Difference Between 0.25 and 0 Gy Comet Measurements on Background Damage V l l l Figure 5.6 Dose-Response Curves for 1 and 3 Hour Rinse Durations Figure 5.7 The Difference in Apparent Damage Dose-Response due to Alkaline Rinse Duration Figure 5.8 Dose-Response Curves for Rinse Durations of 1,3 and 20 Hours Figure 5.9 The Effect of EDTA in Alkaline Rinse on Apparent Damage ix Acknowledgements Acknowledgements are due to the following (at least): my laboratory co-workers, especially Judit Banath and Susan MacPhail, for helpful hints on laboratory technique; Ralph Durand, for helpful discussions on image analysis and software; but mostly Peggy Olive, for helpful advice, specific and general, and for patience. 1 1. Introduction 1.1 The Significance of DNA Damage Ionizing radiation is effective at killing tissue and is used in radiation therapy of cancer to kill tumour tissue. This killing action is due to induced DNA damage occurring either directly, by ionization of the DNA molecule, or indirectly, by generation of reactive chemical species from the surrounding solvent which interact with the DNA molecule. A large variety of DNA lesions is induced by radiation; of particular interest are double strand breaks (dsbs), induced at an approximate rate of 40 per cell per Gy, single-strand breaks (ssbs), at 500-1000 per cell per Gy, and alkali-labile sites (sites which are converted to single-strand breaks in the presence of alkali) at 200-300 per cell per Gy (Powell and McMillan, 1990). While ionizing radiation causes damage to structures throughout the cell, its killing effect appears to be due to damage to nuclear DNA that leads to lethal chromosomal aberrations. This view is supported by the close correlations between cell killing and chromosomal aberrations, and double-strand break induction and chromosomal aberrations (Iliakis, 1991). In contrast, there is a lack of evidence that ssbs are involved in cell killing by radiation, since large amounts of ssbs induced chemically with hydrogen peroxide do not result in high cell-killing (Frankenburg-Schwager, 1990). While ssbs do not appear to be important in cell killing by radiation, they are more easily detected at levels relevant to human radiation exposures. Single strand damage may also represent a lesion of importance in other areas of general biology. DNA damage is a universal problem for life. While genomic integrity is essential for the existence of animal species, the molecules of DNA responsible for this transmission of information between generations of animals are of ordinary chemical stability and suffer all the chemical damage produced in a warm aqueous environment. Many types of DNA damage occur spontaneously from two dominant sources: thermal energy at physiologic temperatures and reactive oxygen species released as by-products of cellular respiration. 2 These sources of damage are expected to establish a background level of damage which may be important when measuring low levels of radiation-induced damage. The overall background rate of DNA damage of all types occurring in mammalian cells, expressed per cell per day, is estimated at 72,000 of which 69,000 are ssbs and only 8.8 are dsbs (Tice and Setlow, 1985; Bernstein & Bernstein, 1991). In terms of the damage induced by ionizing radiation, this number of ssbs would result from 36 to 72 Gy/day and the dsbs from 0.22 Gy/day of x-rays. This level of damage is considerable, as demonstrated by comparison to the L D 5 0 / 3 0 for humans (the dose which yields 50% lethality in 30 days) of 6 Sv (Bacq and Alexander, 1961). These rates of background damage for ssbs and dsbs correspond, respectively, to approximately 600 to 1200% and 7% of an L D 5 0 / 3 0 exposure per day. While the dsb induction rate is apparently the most relevant for radiation killing of rapidly cycling cells such as cancer cells, ssbs or DNA-protein crosslinks may be involved in the ageing process of post-mitotic tissues by interfering with effective messenger RNA production and associated changes to the signal transduction pathways (Tice and Setlow, 1985; and Bernstein & Bernstein, 1991). Single-strand breaks appear to accumulate in post-mitotic or slowly-cycling cell populations such as liver and muscle but are not universally observed; in brain tissue, the evidence overwhelmingly supports single-strand break accumulations with age, with levels of accumulated single-strand damage equivalent to several Gy of x-ray induced damage (Bernstein & Bernstein, 1991; and Rao and Loeb, 1992). 1.2 Overview of the Alkaline Comet Assay The alkaline comet assay described here is a sensitive technique for measuring single-strand breaks developed by Olive et al. (1990) based on the original method of Ostling and Johanson (1984). Briefly, a single-cell suspension is mixed with agarose and pipetted onto a microscope slide. After the gel sets, the slide is transferred to an alkaline, high-salt lysis solution where the cellular protein and lipids are removed and the DNA and 3 RNA are left embedded in the agarose gel. Under alkaline conditions, the complementary DNA strands are free to denature and unwind, but are likely constrained to some degree by the gel matrix. The slides are placed in an alkaline rinsing solution where the salt is removed, then placed in a horizontal electrophoresis chamber. After alkaline electrophoresis, the slides are rinsed in distilled water, followed by staining with a fluorescent DNA-binding dye, propidium iodide (PI). The resulting image has the appearance of a celestial comet (Figure 1.1); hence, "the comet assay". The comet consists of a circular area of bright DNA (the head), surrounded by DNA that has expanded out of the head in a circular pattern (the halo), and the DNA which has streamed out of the head during electrophoresis (the tail). The exact nature of the comet tail is not clear but likely is comprised of loops or strands of DNA which remain connected to the comet head, since the average length of DNA is large at the low radiation doses considered here (0 to 1 Gy). (The average DNA strand length after a dose of 1 Gy x-rays is approximately 1000 u.m, ten times the usual tail length.) The original assay developed by Ostling and Johansen (1984), and similar assays using a non-denaturing pH during electrophoresis (Muller et al., 1994), appear to be sensitive to single-strand breaks due to the extension of loops of double-stranded DNA supercoils during electrophoresis after relaxation of supercoiling by single-strand breaks. The formation of the comet tail is generally accepted to be due to the migration of separated strands at high damage levels, while at low damage levels, the tail is formed by the stretching of strands or loops which remain attached to the comet head (Fairbairn etal, 1995). After staining, the slides are placed in cold storage until measurements are made. The images of comets are obtained using a fluorescent microscope and digital CCD camera. Image analysis is performed using custom software operating on a personal computer. The amount of DNA comprising the comet is approximated by the total fluorescence of the comet image, which roughly distinguishes the cell-cycle position of the cell which originated the comet (ie. Gl , S and G2). Several other comet image aspects are HALO MEAD Figure 1.1 F luorescent image of a 1 Gy comet . The comet was produced with the s tandard assay technique after 1 Gy i r rad ia t ion. The stain was 2.5 / x g / m L of PI in 0.02 M NaCl . 5 calculated (described below in Section 3.4 Definition of Comet Values); the most important are the tail length and the tail moment. The tail length is approximately the distance from the leading edge of the tail to the edge of the comet head, and the tail moment is the distance from the centre of the comet head to the centre of the comet tail distribution multiplied by the fraction of fluorescence in the tail. These two measurements are the features used to quantitate the molecular weight of the comet DNA; by comparison to untreated, control comets, they are measures of DNA damage. The sensitivity of the assay is taken as the lowest dose of radiation necessary to produce comets of significantly higher damage level than controls, based on the statistical significance of the difference in population mean values. There are two aspects limiting the assay sensitivity, the width of the distribution of damage measurements for a single sample (the resolution between populations of similar damage) and the linearity of the measurement response to DNA damage (the amount of variation or scatter for samples of the same induced damage). While the accuracy of a population mean can be ever-increased by increasing the number of measurements for each sample, the scatter (perhaps due to background damage) cannot be decreased in this way. The work described below investigated the factors influencing detection of low levels of DNA damage induced by ionizing radiation in three sections. The resolution of damage in similar comets was addressed in the first two sections: factors involving the measurement of the comet image, largely consisting of software design; and factors influencing the detection of the comet image from the background, largely concerned with DNA staining technique. The third and final section examined the origin of background damage contributing to variability between identical samples, and the effect of variations in the assay protocol. 6 2. General Methods and Materials The handling of cells, irradiation conditions and assay solutions were common to all experiments except where noted. 2.1 Cell Culture and Handling Unless otherwise stated, all experiments used plateau-phase mouse fibroblast cells (C3H 10T!/2) grown in culture. This cell line was obtained from Dr. C. Heidelberger, University of Wisconsin, and cultures were initiated from stocks frozen at passage 8. Some experiments used asynchronous C h i n e s e hamster lung fibroblast cells (V79-171b) grown in culture. This cell line was obtained from Dr. R; Durand, B.C. Cancer Research Centre. Plateau-phase C3H IOTV2 cells were obtained after growing in culture for more than fourteen days after seeding with one-quarter of the number of cells obtained from a confluent plate. They were recognized as plateau-phase by the confluent appearance of the cells in culture and the absence of S-phase cells determined from comet fluorescence measurements. Culture medium (minimum essential medium with 10% fetal bovine serum and 1% antimycotic) was changed weekly for cell culture plates of C3H lOTYi. Cell suspensions were obtained by 10 minute trypsin treatment (0.1% trypsin in citrate buffer), followed by resuspension in 10 mL suspension medium (minimum essential medium with 10% fetal bovine serum) and dilution to approximately 10,000 cells per mL for irradiation and subsequent mixing with agarose. Unless otherwise noted, all experiments used cell suspensions in this suspension medium. 2.2 Sample Irradiation Conditions and Radiation Dosimetry Cell suspensions were irradiated in 5 mL plastic tubes immersed in an ice-water filled jig. Irradiation of samples was performed using the Cs 1 3 7 unit at the BC Cancer Research Centre. Dosimetry measurements were made using a Victoreen Model 500 7 Electrometer with a 0.6 cm3 ionization chamber of wall thickness 0.5 mm. Measurements were taken with the probe at the sample irradiation position in the jig. To ensure charged-particle equilibrium, a rubber balloon was placed tightly over the probe and the probe was placed in a water-filled 5 mL tube in the water-filled radiation jig. The dosimetry calculations were performed as described by Johns and Cunningham (pp. 246-8, 1983). The dose to the medium at the position of the probe with the probe removed, D m e d , is given by Dmed = M N x (0.00873 J/kg R) ( u ^ p ) ™ ^ where M is the electrometer reading in roentgens (R), N x is a unitless calibration factor (equal to the exposure in R divided by the electrometer reading in R), and ( H a t / P ) m e d a i r 1 S m e mass-energy absorption coefficient ratio for the medium and air. From Table 7-4 of Johns and Cunningham (1983), (M.ab/p)meuair= 1.112 (unitless) for the medium being water. The electrometer calibration factor N x was reported to be 1.12226 for a 250 kVp beam (from notes accompanying the electrometer by Kornelsen and Skarsgard, 87/1/14), but a calibration factor was not available for Cs 1 3 7 mean photon energy of 662 keV. This calibration factor was thus modified to reflect the different response of the electrometer for different energy photons. The response of the electrometer-probe assembly varied slightly with photon energy. Correction factors were reported (according to electrometer documentation by Nuclear Associates Inc.) to be 0.992, 0.996 and 1.005 for 200 kVp, 280 kVp and Co 6 0, respectively, where the true exposure is equal to the exposure reading multiplied by the correction factor. Interpolating, the correction factor at 250 kVp and Cs 1 3 7 mean photon energy are found to be 0.995 and 1.000. The calibration factor for the electrometer at Cs 1 3 7 mean photon energy was taken as the calibration factor at 250 kVp multiplied by (1.000/0.995). Hence, N x for Cs 1 3 7 photons is 1.1279. The electrometer reading M was corrected from the actual measured reading M' by the temperature T and pressure P: M = M' ( T [K] / 293 )( 760 / P [mmHg] ) = 1.00381 M* 8 with water bath temperature of 21°C (294 K) and air pressure of 759.7 mmHg (from Princo Barometer, corrected for temperature and latitude). Hence, the dose to the water at the position of the probe was found to be: Dmed = M ' (1-00381)(1.1279) (0.00873) (1.112) Gy = 0.01099 M' Gy. The exposure reading and residuals (difference between the values predicted from a linear regression and the experimental data) are shown in Figure 2.1. Both dose rates used the Cs 1 3 7 unit with a lead attenuator in place to reduce the dose rate. The high dose rate was obtained with the jig mounted on extensions. The low dose rate was obtained with the jig set across the radiation room on a set of mobile shelves. Electrometer readings were taken for ten time settings for each dose rate. The dose to water at the jig position, D m e d , was calculated from the equation above, with linear coefficients and shutter timing-errors as indicated in Table 2.1. For an unknown reason, one data point for the high dose rate had dramatically higher residual, as seen in Figure 2.1, and was excluded from the dosimetry calculations. Since the shutter-timing error must be independent of the jig position, the shutter timing-error was taken to be the average of these values, 0.043 min. The error in the shutter timing-error was taken to be of the order 0.004 minutes, a value slightly larger than the stated errors obtained from linear regression due to the deviations between these values. The irradiation times used for cell samples were always greater than 0.200 min, yielding a maximum error in the delivered dose of approximately 2.5% due to the combined errors in shutter-timing error and dose rate. 9 E x p o s u r e T ime (m in ) F igu re 2 . 1 E l e c t r o m e t e r r ead ings f o r d i f fe ren t e x p o s u r e t i m e s . The res idua l is the d i f f e rence be tween the f i t ted l ine and the m e a s u r e d v a l u e s . 10 Table 2.1 Dosimetry Values Dose Rate Shutter-Timing Error Dose Rate (minutes) (Gy/minute) Low +0.046±4.3% 0.0500±0.13 % Medium +0.039±1.9% 0.311±0.057% High +0.044±15% 6.97±1.0% High +0.044±4.7%* 6.94±0.31%* * one data point of unusually high deviation from linear fit was excluded from analysis. 11 2.3 Comet Assay Technique After irradiation, 5 mL plastic tubes were prepared, each containing 0.5 mL of cell suspension. For experiments which used single samples for each radiation dose, these 0.5 mL samples were prepared before irradiation and irradiated individually. For experiments using many samples for each dose, a large quantity (up to 4 mL) was irradiated, then split into 0.5 mL quantities. 1.5 mL of 1% w/v agarose in PBS (at 40°C except where noted) was added to each tube using a 5 mL pipette, mixed by repeatedly pipetting the solution; and 1.5 mL of the solution was pipetted onto a microscope slide situated on an inverted, ice-filled glass tray. The microscope slides were frosted on one end and large drops of agarose solution had been dried to both ends of the slide to assist in anchoring the agarose gel to the glass. The time from the addition of agarose solution to the cell sample to the pipetting of the gel onto the slide was typically 10 seconds. The gel was allowed to cool for a minimum of 2 minutes before transfer to lysis solution. Lysis solution consisted of 0.03 M NaOH, 1.2 M NaCl and 0.1% N-lauroylsarcosine (prepared by mixing 400 mL double-distilled water, 27.8 g NaCl, 0.48 g NaOH, and 0.4 g N-lauroylsarcosine). Some experiments used 1.0 M NaCl instead; the solution was later changed to 1.2 M NaCl since other workers found evidence that removal of histones was not complete in 1.0 M NaCl and may influence damage measurements (Olive and Banath, 1995). A maximum of 10 slides were treated with each 400 mL batch of lysis solution. Since the slides must be prepared sequentially, the time spent in one of the steps must differ slightly for each slide in a batch of slides which were electrophoresed together. To address this timing problem, the lysis step was extended for the slides prepared first since the lysis duration was found to have least effect on comet measurements. Typical batches of slides required approximately 1 minute per slide to prepare, so the lysis time would be extended by 1 minute per slide in the set of slides. After treatment in lysis solution for 1 hour, the slides were transferred to alkaline rinse solution (0.03 M NaOH, 2 mM EDTA) on plastic racks of 4 or 5 slides per rack. The slides were left for a period of 12 slightly less than 1 or 3 hours then transferred to a flat electrophoresis chamber containing alkaline rinse solution as above. The slides were electrophoresed at 0.6 V/cm for 24 minutes starting at 1 hour after the end of the lysis period (for an accurate alkaline rinse duration of 60 minutes). After electrophoresis the slides were transferred promptly back to plastic racks and placed in pure distilled water for a minimum of 30 minutes. After this rinse the slides were placed in a staining solution of 0.02 M NaCl and propidium iodide (PI). The concentration of PI varied between experiments. After staining, the slides were transferred to sealed plastic containers lined with paper towels dampened with distilled water to prevent drying of the agarose. Slides were typically stained overnight. The concentration of PI was often 0.025 u.g/mL but sometimes 0.25 ug/mL. Some experiments used a 30 minute stain in 2.5 u.g/mL PI. (The effect of staining time and concentration of PI and NaCl is examined in Chapter 4.) The main effect of decreasing dye concentration was to decrease background fluorescence; there was little effect on comet measurements. Based on studies of dye binding to DNA, minimization of dye concentration was expected to yield increased resolution of the comet image, prompting a strategy of using decreased dye concentration. However, at very low dye concentrations some slides appeared unusually dim while other slides in the same set were normal. These dim slides may have been affected by contacting the distilled water from paper towels or condensation during storage. Comets were usually measured within two days of staining. The slides were placed on a fluorescent microscope with a CCD camera and personal computer attached. For routine comet measurements, the camera was allowed to operate for several hours for the response to become stable (see Chapter 3). Measurements were written to a floppy disk for later analysis. Fifty comets were measured for cell samples of single populations (plateau-phase C3H 10T1/2 cells) and 100 or more for samples of more than one population (cycling V79 cells). 13 2.4 Statistical Analysis The significance of the difference in means between two samples was determined using the two sample t-test assuming unequal population variances. The significance of the difference in sample variances was determined using the F-test. For linear regression, the coefficient of determination, r2 or R, was taken to indicate the goodness of fit (the proportion of observed variation explained by the linear regression model). The significance of the difference from zero of the slope of the regression is given by the F test (which is the same as one-way ANOVA analysis). The t-test was performed on data using custom software utilizing Numerical Recipes in C routines (Press et al., 1992). F-test, r 2 values and regression parameters were obtained using SPSS for Windows Student Version Release 6.0.1 software or Sigma Plot version 5.0, Jandel Corporation. Errors in values resulting from algebraic manipulations were approximated by error propagation using the standard conventions of addition of fractional errors when multiplying or dividing, and addition of absolute errors when summing or subtracting. 14 3. Development of Software for Comet Analysis 3.1 Introduction and Objectives 3.1.1 Design Objectives There were two objectives for writing new software for comet analysis. The first objective was to produce additional information about the comet image that is not provided by the established software; this additional information, such as background fluorescence, was necessary for examining factors influencing the assay sensitivity due to factors other than the direct measure of damage. For example, by measuring background fluorescence, staining technique could be studied to optimize the detection of the comet image. The second objective was to improve sensitivity of the image analysis for low-damage comets by improving the algorithm used to discriminate between the comet image and the background. The software that was developed provided a range of measurements on the comet image. These include the familiar tail length, tail moment, total fluorescence (DNA content), percent fluorescence (DNA) in tail, and area of the total comet. Additional measurements are made of the area of the tail, width of the comet head and halo, total screen fluorescence, background fluorescence, and additional measurements of the tail moment and comet fluorescence that use slightly different calculation methods. These redundant measurements were intended for determining experimental error during the measurement process. 3.1.2 Model of Camera Response As with the established comet analysis software, the new software utilized the digitized image of the fluorescent comet, which was passed to a computer memory board from a CCD camera mounted on the fluorescent microscope. A smaller area of 100x600 camera elements, assumed to contain one comet, was converted into a floating-point matrix 15 of 50x120 elements, the image matrix, which formed the basis for all comet calculations. This screen area was located close to the screen centre and was large enough to contain one comet of usual proportions. This eliminates much of the debris that would interfere with comet recognition, and reduces the effects of the varying camera response across the field of view. The image matrix uses the average of a 5x5 square of camera element values per floating-point element. This conversion into floating-point was done with or without processing for uneven camera response as described below. Image analysis is based on a model of the relationship between this digitized image and the fluorescent pattern of the comet in the gel. The camera signal was assumed to be linear with an offset with respect to the light striking the camera at a particular element. The camera gain setting was adjusted to the maximum setting that did not saturate the detector; the camera provided a range of approximately 256 on the image board for each element. The offset represents the dark current, the camera image obtained when no light is striking the camera sensors. The dark current of a CCD camera image is generally regarded as due to thermal excitation of the CCD detector, and follows the diode law relation between current and temperature (Mackay, 1986), but may also involve contributions from the computer memory board or other sources. The camera response for element / is modelled in the form, ' S^afr + b, (1) where St is the camera output signal, F, is the light reaching the area i of the camera detector, a, is a camera linearity constant and bt is an offset value. The fluorescent light emitted from the comet is assumed to be proportional to the amount of fluorescing dye present, At, and the amount of illuminating light, It: FrA,I, (2) The fluorescence characteristics of a dye will depend on a variety of factors, such as binding mode of the dye to the site of binding on the DNA; but the fluorescence of the particular 16 dye molecule (fluorophore) should be linear with illumination, characterized by the quantum yield. Combining (1) and (2) gives: S^a^ + b, (3). From this relation, image element offset values bj can be found from the dark current image (the dark-field, which is the signal obtained from the camera with the shutter closed); ie. with /, = 0, S, = bt. Concerning the uniformity of the CCD response, at, "The only uniform CCD is a dead CCD" (Mackay, 1986). Following a procedure developed for astronomical CCD camera use, a flat-field (uniformly fluorescent image) was used to calibrate the response of the different elements to compensate for non-uniformities of response in a-v Rearranging (3) yields: IflrfSfbJ/At (4) For a flat-field, At will be constant for all elements, assigned AB; and the signal will be Sm for element /. From this condition, a relationship can be found for the camera linearity constants and illumination levels: AB=(SBi - bt)/Ifll=(SBk - bk)/Ipk (5) for elements / and k. Rearranging (5) yields: IflrfSm-bOlM/fin-bt) (6) This relation potentially allows the software to compensate for uneven illumination of the field of view or for varying camera linearity constants. Without an independent measurement of It or a,, values of /, and a, cannot be determined separately. Since absolute values of fluorescence are not required for the measurements involved in the comet assay, these values do not need to be separated. The desired measurement from each element is the amount of DNA present, as indicated by the fluorescence of the dye-DNA complex. From (4) and (6), A,=(S, - bj/lfr = (S, - bt)(SBk - bk)/ ftft (Sm - b,)) (7) 17 This last relation depends only on values that can be obtained from the camera signal, except for the ljfxk which cannot be found but acts as a normalizing factor. It is desirable to choose an arbitrary value for (SkB - bk)/Ikak such that the data obtained are easily comparable whether or not software compensation is used. A convenient requirement is for the total amount of fluorescence measured over the entire image for the blank, uniformly fluorescent screen to be equal when the camera signal is compensated or not compensated for uneven response. The measured values of AB with compensation, Amc, and without, ABtu, will not be equal to AB due to measurement error. For element i the uncompensated signal, IfOj is taken as constant (set to 1) and T.iAw^tfSn-bt) (8). (All sums run from 1 to N, the number of elements in the image). For the compensated signal from (7), E, ABUC = Z, (SBi - bt)/Ifll = E, (SBi - bt)(SBk - bk)/ (1^ (SBi - b,)) = ((sBk-hMW 2, (Sm - b,)/(Sm - bt) -(Su-btJN/fa (9) Equating the expressions for the corrected and uncorrected signals, (8) and (9), yields: (SBk-bk)/Ikak = 2Zi(SBi-bi)/N (10) which is the mean of the blank-view, camera element signals. Finally, from (7) and (10) the corrected image intensity measured by camera element / is A,=(St - bjp, (SBj - bj)l N)/(SBi -bt) (11) after changing indices to avoid confusion. The software utilizes arrays of camera offset values bt and weighting values, wt, wrVjtSsj-bjJ/NMSn-bj) (12), which are stored in an array after being evaluated at the start of, or during, a session of comet measurements. The image intensity at element i is then evaluated as A^fa-b,) (13). 18 Using this model for the amount of DNA represented by a given camera signal, software was written to calculate comet values. 3.2 Methods and Materials All comet measurements were made on the Fluorescent Image Processing System (FIPS), consisting of an intensified solid state CCD camera mounted on a Zeiss epifluorescence microscope. Images were passed from a CCD camera (F4388 generation III gated intensifier camera from ITT, Electro-Optical Products Division, Roanoke, VA) to an ITEX (MFG-IM-V) memory board (supplied by Imaging Technology Inc., Bedford, MA) in a 80486, MS DOS-based personal computer. Software was written in C and compiled using Microsoft Quick C compiler. The ITEX Visionplus library functions were used to read image data for computations. 3.3 Results and Discussion 3.3.1 Experimental Evaluation of Camera Model The above equation (13) converts the camera output image to an image representing the amount of fluorescence that would be present in the sample if camera response and illumination were constant across the field of measurement. However, this analysis assumes that the fluorescence seen by the camera originates at the comet in a clear manner. The intervening gel will absorb and emit light and will bind dye to some extent. The comet also has depth, and will not be in focus throughout this depth. These limitations, as well as other unidentified ones, will limit the extent to which this type of compensation may be successful. The test of success or failure is based on the ability of the software to produce reasonable data, in terms of comet measurements, and in terms of camera response. One expectation is that the camera response for a flat-field be composed of random deviations from a mean value with a Gaussian distribution of element intensities (Massey & Jacoby, 1992). The flat-field standard consisted of a microscope slide with adhesive tape 19 placed on the underside. The tape produced a strong fluorescent image but had to be placed out of focus to suppress surface details. Other flat-field standards were tried, such as propidium iodide in solution or dried onto a slide; but these faded quickly and had defects such as large crystal patterns. The distributions of image values for images of low and high flat-field illumination are shown in Figures 3.1 and 3.2 using each compensation method. (Note that in the following comparisons, flat-field weighting always includes dark-field subtraction, since flat-field weighting requires dark-field subtraction in calculating the weighting values.) The effect of dark-current subtraction and flat-field correction is to narrow the distribution, improve the symmetry, and shift the distribution by the amount of the mean offset. These results are seen for both low illumination and high illumination, though the image intensity at the higher illumination is still only at approximately 20 on a range of 0 to -240 (saturation appeared to begin near an element value of 240, judging by the maximum comet element measured for a changing gain setting). Values outlying the distribution are present for the flat-field corrected image. The linearity of camera response was measured indirectly. A comet with large tail (5 Gy damage) was measured under high and low illumination The different illumination levels were expected to produce elements whose values at high and low illumination are of the same proportion, limited by the uniformity of the filters used to set the illumination level. Plotting each element on a graph according to its signal at high and low illumination should then produce a line, the slope being the ratio of illumination intensities (shown for each image processing method in Figure 3.3 ). The linearity, as indicated by the coefficient of determination, r2, is highest for the dark-field subtraction compensation without flat-field weighting. Again, outlying elements are seen for the image compensated with flat-field weighting as in Figure 3.1 and 3.2. The camera response was found to change with time, especially during the first two hours of use. Figure 3.4 shows the dark-field image for a cold and warm camera. The warm camera had an increased dark-field mean intensity as well as an altered image 20 CO c CD E _CD LU CD jQ E Z3 1400 1 2 0 0 1 0 0 0 8 0 0 600 4 0 0 2 0 0 0 h 1400 1200 1000 8 0 0 6 0 0 h 4 0 0 2 0 0 -0 -_ - 1 0 Compensa ted using dark—field subtract ion C o m p e n s a t e d using flat—field weighting - 5 0 10 15 2 0 2 5 30 F l u o r e s c e n c e (arb i t rary uni ts) Figure 3.1 The distr ibution of f luorescence for a low i l luminat ion, blank field of view. The effect of dark—field subt ract ion and flat—field weighting is to narrow the d i s t r i -but ion. The flat—field weighting also increases the number of outlying e lements. 21 CO •+-> rz CD E _CD LU CD _Q E 1 8 0 0 1600 1400 1200 1 0 0 0 8 0 0 6 0 0 4 0 0 2 0 0 0 1800 1600 1400 1200 1 0 0 0 8 0 0 600 4 0 0 -2 0 0 -0 fa Compensated using dark—field subtraction • / / Compensated using flat—field weighting - 2 0 0 20 4 0 60 80 100 Fluorescence (arbitrary units) Figure 3.2 The distr ibution of f luorescence for a high i l luminat ion, blank field of view. The effect of dark—field subt ract ion and flat—field weighting is to narrow the d i s t r i -but ion. The flat—field weighting also increases the number of outlying elements. 22 70 10 50 100 150 2 0 0 2 5 0 100 150 2 0 0 2 5 0 0 50 100 150 2 0 0 2 5 0 High E lemen t Va lue a) no compensa t ion r 2 = 0 . 9 8 4 in te rcep t=4 .58 b) d a r k - f i e l d c) f l a t - f i e l d weighting sub t rac t ion r 2 = 0 . 9 8 9 1^=0.991 i n t e r c e p t = - 0 . 1 91 i n t e r c e p t = - 0 . 1 88 Figure 3.3 The l inear i ty of c a m e r a r e s p o n s e . The l inear i ty of the c a m e r a r e s p o n s e is d e m o n s t r a t e d by p lo t t ing e a c h e lemen t of c o m e t i m a g e a c c o r d i n g to i ts va lue at high and low i l l u m i n a t i o n . The i m a g e s were (a) u n c o m p e n s a t e d , c o m p e n s a t e d by (b) dark—f ie ld s u b t r a c t i o n or by (c) f l a t - f i e l d we igh t i ng . The c o e f f i c i e n t of d e t e r m i n a t i o n ( r 2 ) and o f f se t were found f r o m l inear r e g r e s s i o n a re s h o w n . 23 Figure 3 . 4 The d a r k - f i e l d i m a g e fo r a c a m e r a with s h u t t e r c l o s e d , a) Co ld c a m e r a ( recen t l y t u rned —on), b) Warm c a m e r a ( o p e r a t e d fo r seve ra l h o u r s ) . 24 co c CD CD CD D CD E 1000 800 -600 400 200 0 h . JL T 1 1 1 - I 1 1 1 !•• 1 1 1 I 1 10 min H 1000 800 600 400 200 0 T — I — i — r T — i — | — i — i — i — | — r ~ 20 min ; 1 1 ,i 1 1 1 i 60 min 1 1 i 1 1 1 i 100 min i 1 i 1 i 1 i 1 i 140 min -1 0 1 2 3 4 5 6 7 8 9 1 0 1 1 Element Value F igure 3.5 The D is t r ibu t ion of d a r k - f i e l d e l e m e n t i n tens i t y with t i m e . Ini t ial ly, the d is t r i bu t i on c h a n g e s rap id ly , c o m i n g to s t eady—s ta te a f te r a p p r o x i m a t e l y two h o u r s . 25 pattern. The change in the distribution of intensities over time is shown in Figure 3.5. This changing response affected the comet data produced by both image analysis programs, shown in Figures 3.6 and 3.7 for the measurement of total fluorescence. Figure 3.7 shows the changing values of DNA content with time for comets from slides corresponding to the data points at 0.5 and 4.5 hours of Figure 3.6. A clear decrease in fluorescence occurs with order of measurement for the data set taken at 0.5 hours. The data set taken at 4.5 hours does not show this decline, eliminating the possibility that the drop is due to bleaching of the slide. All subsequent comet measurements were taken with a warm camera, after two hours with power on to both camera and computer. 3.3.2 Definition of Tail End The comet tail appearance, especially at low damage levels, is very irregular; this makes it difficult to distinguish the comet tail elements from the background noise. An edge-finding algorithm was avoided for this reason, and a global thresholding strategy (as described by Gonzalez & Woods (1992)) was adopted to distinguish comet elements entirely by their values. Global thresholding consists of assigning the image elements to either background or comet image based on their magnitudes below or above a threshold value which is constant across the entire image matrix. Since the background fluorescence may change between comets due to differing depth in the gel and the presence of out-of-focus comets, the threshold is taken as the mean of the background fluorescence (evaluated at the top and bottom rows) plus n standard deviations, where an optimal value for n was determined empirically. This threshold corresponds to the choice of an acceptable probability of misclassifying a background element as a comet element and vice versa. The comet value most sensitive to this choice of global threshold was the tail length, which is defined as the distance from the head edge to the tail end. The tail end is defined as the leftmost image matrix column containing more than / comet elements. The choices of /, the tail end size, and n were determined jointly by successive analysis of 26 18 16 O 14 12 10 8 4 h-0 T r - 2 - 1 0 1 2 3 4 5 6 Duration of Camera Use (hr) 7 - 6 o o 3 ° 5 3 CD CD CO r-+-r-1-4 E E co' o zr , CD 3 C L CO o Q CD 0 7 8 Figure 3.6 The change in comet f luorescence with camera useage. The same slide of comets was measured repeatedly. Each symbol is the mean ± standard deviation for 50 comets . 27 C o m e t N u m b e r F igu re 3.7 C h a n g e s in c o m e t f l u o r e s c e n c e with o r d e r of m e a s u r e m e n t . The c o m e t n u m b e r is the o rde r of m e a s u r e -m e n t . The i n d i c a t e d t i m e is the t ime the se t of m e a s u r e -m e n t s was b e g u n . The c o e f f i c i e n t of d e t e r m i n a t i o n ( r 2 ) is g iven f r o m l inear r e g r e s s i o n . 28 stored comet images using different values of global threshold and tail end size. This method allowed the comparison of an identical comet image for a large number of values. The image of a comet showing slight damage (and a faint tail) and a typical comet produced following 1 Gy radiation damage are shown in Figure 3.8 and 3.9, where each square represents an element above the background mean plus n standard deviations. For increasing n, the comet area is reduced, especially in the tail region. The uncompensated image shows a distinct area in the upper left of the image that is above the threshold (hence part of the comet image) but clearly separate from the proper comet image as observed directly. These elements are referred to as outliers. Dark-field subtraction dramatically reduces the number of these elements. The image compensated using dark-field subtraction has a much higher background to the left of the image, while the image compensated using flat-field weighting has a more even background. An uneven illumination was the apparent cause, since later observations showed a considerable variation in comet intensity when measured at different positions on the screen. These results suggest that the use of flat-field weightings successfully compensates for uneven illumination. However, these corrections were not found to be reliable; several elements were consistently weighted too heavily, causing inappropriate tail ends to be found. The images obtained using dark-field subtraction at three standard deviations and flat-field weighting at two standard deviations have very similar outlines of the comet shape; but the outlying elements are far more numerous for the image compensated using flat-field weighting. The threshold required to eliminate background noise for the uncompensated image discards nearly the entire tail area; and analysis using the uncompensated image would fail to find the tail end due to the noise at the left edge of the screen. The joint effect of tail end size and threshold number of standard deviations is shown in Figure 3.10 for a highly-damaged 1 Gy comet. For zero threshold all choices of tail end size cause the tail end to be found at the left edge of the screen. As the threshold q o —\ m -q 6Z q O -H r-t- i-t-• z r _ . x CD q C D O O 3 CD r-t-Q CO O 3 cr Q o l O -1 o c C L rz o -1 CD CO O CD Z5 o CD CD CD 3 — CD Q CO CO Q C 3 5 "0_ CL CD ~ CO CD 3 CD Q ZJ CO <-t-z r CD Q ZT CD Q r-l-r+- O CD XJ C L C L CT O r-t-(-+-O 3 -i o CO ZJ C L z r Q C L C L co' ZT CD 5-3 CD O o r-t-CD Q. ZJ C 3 CT CD -\ O —h CO r-t-Q 13 C L Q CD < o ZJ CO m 3] Q 0 z r ur CO CD OJ c Q CD - i CD 1 -1 1 z r CD CD T3 —s O es c l-t-CD Z5 Z5* r-f-co CD Q O -+1 Q 3 Q IQ CD CD CD O CD O me 3 CD r-t-ZJ r-l- O c r 0 z r < O CD CO CD 1-1-z r < CD Q lue DUJj . IQ CO CD Q r-f-c r z r 0 < CD CD CO z r r-t- o_ ZT CD C L im Q CD r-t-ZT -1 CD CO z r 0 C L 3 O O 3 T J CD 3 CO a i-t-CD C L D Q -1 TT I 31 cp_" C L CO c cr Q O r-t-o' 3 31 Q C L CD U D * 3 " !-t-3 * oe 31 Tail End S ize : 120 100 80 60 40 20 0 120 100 80 60 40 E 20 =i 0 120 CN 100 X 80 N ' 60 J Z - M 40 C7> c 20 Le 0 120 100 I— 80 60 40 20 0 120 100 80 60 40 20 V \ *; _ \ j \ i \ V * — i ; ~~——— • X — k i \ \ t \ \ \ \ \ \ — — — -\ ; \ \ V ; \ i - _ : \ -< \ X Threshold (number of SD above mean) Figure 3.10 The effect of tail end size and threshold on tail length for a 1 Gy comet. Tail lengths with no compensat ion , dark —field subtract ion , and f l a t - f i e l d weighting — - • are shown for thresholds given as the number of standard deviations (SD) above the mean background for tail end s izes shown. 32 increases, the calculated value of tail length is reduced quickly until the value of tail end is reached that corresponds to the expected position of the tail end, as observed directly; and a plateau with small slope occurs. As the threshold increases, the tail length is further reduced. The profiles for the three methods of image analysis are shown. The image with dark-field subtraction has the largest plateau for all tail end sizes; the image with flat-field weighting has the next largest plateau; and the uncompensated image has only a small plateau, indicating that it is very poor at finding the tail end. The same trends appear for absolute threshold as shown in Figure 3.11. The tail length versus threshold profiles for a comet with a faint tail are shown in Figure 3.12 and 3.13; these show similar characteristics as above. Increasing the threshold to four standard deviations discards a greater amount of the comet image. Based on these observations, routine comet measurements were made with a threshold equal to the background mean plus three standard deviations and a tail end size of three. This choice was made due to the large plateau and sharper edges seen in the plateau of Figure 3.10 and 3.12 for tail end size of three. The large plateau for tail end size indicates that the tail length is relatively insensitive to the threshold. The choice of three standard deviations for the threshold number was made since this marks the lower limit for identifying the tail end, thus minimizing the number of comet elements excluded. Dark-field subtraction was used; flat-field weighting was not used due to the generation of outlying elements which masked the tail end. 3.4 Definition of Comet Values The software calculates other attributes of the comet for file export and for calculation purposes. The edges of the comet head are found from the positions of maximum or minimum derivative with respect to horizontal distance, which is a standard edge-finding technique (Gonzalez and Wood, 1992). The left and right edges are found 33 120 100 80 60 40 20 0 120 100 80 60 40 20 0 120 100 80 60 40 20 0 120 100 80 60 40 20 0 120 100 80 60 40 20 0 Tail End Size: 1 1 1 . 1 . 1 1 . —•—•—i—i N - r ~ v - - \ \ - V i \ T \ i \ * e="=j — : s. • - 4 \ \ 2 1 ----^  \ \ —1— 4 \ 1 2 3 4 5 6 7 Absolute Threshold Above Mean Figure 3.11 The effect o f - ta i l ' end size and threshold on tail length for a 1 Gy comet. Tail lengths with no compensat ion — - , da rk - f i e l d subtract ion , and flat—field weighting are shown for thresholds given as absolute values above the background mean for tail end s izes shown. 34 80 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 Tail End Size I i i ' i I ! \ > U % I • — H H T - H H r i . i 1 1 1 1 I _1 — -s s _ - \ \ . \_ 1 w \ S 1 s \\ | \ 0 . 1 2 3 4 5 6 7 8 Threshold (number of SD above mean) Figure 3.12 The effect of tail end size and threshold on tail length for low damage comet . Tail lengths with no compensat ion -, dark—field subtract ion , and f l a t - f i e l d weighting are shown for thresholds given as a number of standard deviat ions (SD) above the background mean. 35 80 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 80 70 60 50 40 30 20 10 0 Tail End S ize : 1 —TT \\ s \ = - — -\ • \ -\ • -r ™ 1 2 3 4 5 6 7 8 Absolute Threshold Above Mean Figure 3.13 The effect of tail end size and threshold on tail length for a low damage comet. Tail lengths with no compensat ion — - , da rk - f i e l d subtract ion - - — , and flat—field weighting are shown for thresholds given ds absolute values above the background mean. 36 from the positions of the maximum and minimum derivatives using image data along the horizontal lines averaged from positions one above, below, and at the mean of the distribution of the fluorescence in the vertical screen direction. The head boundaries are often irregular, especially for plateau-phase cells. (This irregularity in comet appearance disappeared after adopting a protocol using higher temperature agarose and overnight staining duration. See Chapters 4 and 5.) The right halo edge is the right-most image column containing three comet elements, similar to the definition of the tail end. The tail length is taken as the distance from the left head edge to the tail end with the right halo width subtracted. Thus, the tail length is effectively the distance from the left halo edge to the tail end for a symmetric halo (Figure 3.14). This definition is taken since the left halo width would otherwise be measured as tail length for comets lacking tails. The halo width is a function of DNA damage with low-dose comets having large halo width and high-dose comets having a small halo width. The raw tail moment is the sum of the products of comet element values and their distance to the left head edge, divided by the total comet fluorescence: MR = Ifi, (i - iMj /(E a l l,.E a l l, Ay) (14) with horizontal element index / running from the leftmost screen edge to the left head edge, iLH and the index j running from the top to bottom of the image. (A rectangular, two index notation replaces the single index notation used above for A.) As for the tail length measure described above, the zeroed tail moment corrects the tail moment for the expected contribution of the left halo width, assuming the left halo would be the same size as the right halo for a symmetric comet showing no damage. The zeroed tail moment is calculated as: Mz = [EyE; (i - iLH)Ay - XjZk (k - W ^ M S a i i A i i / Av )' with i as above and k running from the right head edge to the right halo edge. The tail moment is the same (to a constant) as the distance from the head edge to the mean of the 37 Head Halo a - tail end b - left head edge c - right head edge d - right halo edge e - left halo edge he ad width = B right halo width = C left halo width = D tail length = A - C Figure 3.14 Diagram of comet measu remen ts . The comet image e lements are determined f rom a global threshold . The edges of the comet head are determined f rom posi t ions of max imum and min imum hor izontal der ivat ive. Edges of the comet halo and tail end are found f rom the number of comet e lements in the co lumn as descr ibed in the text. 38 tail fluorescence distribution multiplied by the percentage fluorescence in the tail. The comet area is taken as the total number of comet elements multiplied by the area of an element; the tail area is the total number of comet elements to the left of the left head edge, reduced by the number of elements between the right head edge and right halo, multiplied by the area of an element. The background fluorescence is measured at the screen top- and bottom-most rows; the mean and standard deviation of each screen is recorded for each comet. The tail fluorescence is the sum of all comet elements to the left of the left head edge, reduced by the fluorescence in the image columns from the right head edge to the right halo edge. The percent of fluorescence in the tail is the tail fluorescence divided by the total comet fluorescence multiplied by 100. The fluorescence of each comet was measured in two ways. The sum of all comet element values is referred to as the comet fluorescence. The value obtained by summing all screen element values and subtracting the background mean is referred to as the expected comet fluorescence. These two measures are generally in close agreement, with the expected comet fluorescence typically slightly higher. A large difference between the two suggests that the chosen threshold has classified much of the comet as background and represents a means of determining the approximate extent of the comet signal discarded. It is not clear exactly what the element intensity represents, since the fluorescent light striking the CCD camera may come from various sources: the stain associated with the comet DNA, the stain in solution or bound to the gel, and glare from various sources. Most significant is the question, does the fluorescence seen as the comet image include the fluorescence produced in the gel before and behind the comet plane of focus? If the measured comet fluorescence has included the background fluorescence due to dye in solution or bound to the agarose gel, then the background should be subtracted to obtain the proper value. Alternatively, the background fluorescence may indicate the comet's depth in the gel, which may absorb the comet fluorescence rather than add to it. The correlation was 39 found to be low between the comet fluorescence and the background fluorescence for thick and thin gels of comets of plateau-phase cells, irradiated with 1 Gy. The coefficients of determination, r2, between the comet fluorescence and background were 0.01 and 0.05 for thin and thick gels respectively, indicating no large contribution of the background to the comet fluorescence. (Data not shown). The dependence of comet fluorescence on the background fluorescence due to free dye in the gel was investigated by producing slides of usual thickness or approximately half the usual thickness by squeezing the gel with another slide before the gel cooled (Table 3.1). The comet fluorescence for thick gels have significantly lower fluorescence (t-test at significance level 0.01) for both radiation doses and independent of background subtraction, suggesting that fluorescence emitted from the comet DNA is partly absorbed by the intervening gel and dye solution. The variability of the comet fluorescence was also significantly reduced for the thick gel (significance of difference in means <0.02 for 0 Gy slides and 0.10 for 1 Gy slides), indicating an improvement in sensitivity for measuring total fluorescence using the thick gel. The variability of the comet fluorescence is independent of the use of background subtraction (differences not significant using F-test for difference in variance at significance level 0.05), indicating that background subtraction does not yield a decreased variability in the measure of DNA content. Since the DNA content of these cells is expected to be the same, the smallest variability is most appropriate and the usual gel preparation was used for subsequent sample preparations. No background subtraction was used for subsequent comet fluorescence measurement. 40 Table 3.1 Measured Fluorescence for Thick and Thin Agarose Gel Dose (Gy) Gel Thickness Calculation Method Fluorescence a 0 thin comet elements 21.21 ±3.17 background subtracted 0 thick comet elements, 19.45 ± 1.35 background subtracted 0 thin comet elements 21.90± 3.18 0 thick comet elements 20.29 ± 1.34 1 thin comet elements, 18.65 ±1.54 background subtracted 1 thick comet elements, 16.86 ± 1.19 background subtracted 1 thin comet elements 19.44 ± 1.49 1 thick comet elements 17.82 ± 1.15 a mean ± standard deviation ot 50 comets tor each slide 3.5 Comparison of New Software with Established Software Measurements made with the new software were compared to measurements made of the same comet samples using the established software. Figure 3.15 shows the results of comet measurements using both programs for x-ray doses of 0 to 30 Gy. Comets were prepared using the standard comet protocol using 1 hour lysis and rinse durations (comet slides were prepared by Judit Banath). The scales of Figure 3.15 are different for measurements made by the different programs. Similar results are obtained for both programs which ensures that no gross error has occurred in the coding of the new algorithm. Measurements using the new and established software were also made for comets of 0 to 1 Gy damage (Figure 3.16) and for asynchronous, unirradiated cells (Figure 3.17). (Note that data from Figures 3.16 and 3.17 were obtained from later experiments and used a 3 hour rinse duration, while Figure 3.15 used a 1 hour rinse duration.) 41 10 c CD E o E o 1 ^ 0.1 J 1_ Dose (Gy) Figure 3.15 Compar ison of comet measurements using the new and establ ished software for high damage. The same set of sl ides for doses of 0 to 30 Gy were measured separately using the software descr ibed in the text o and the establ ished software • . Each symbol represents the mean ± standard deviat ion for 50 comets per s l ide. Note the logar i thmic sca le used on some of the sca les . The units are arbitrary. 42 12 •+-> CD 10 c CD o E $ 8 o 4— ^ o CO 6 o h-(ne 4 2 0 120 ® 100 p CO 80 _ 60 5 5 I - <u c 40 20 20 40 60 80 100 120 D o s e ( c G y ) F i g u r e 3 . 1 6 . C o m p a r i s o n o f c o m e t m e a s u r e m e n t s u s i n g t h e new a n d e s t a b l i s h e d s o f t w a r e f o r low d a m a g e . T h e s a m e s e t o f s l i d e s f o r d o s e s o f 0 t o 1 Gy ( 1 0 0 c G y ) w e r e m e a s u r e d s e p a r a t e l y u s i n g t h e s o f t w a r e d e s c r i b e d in t h e t e x t o a n d t h e e s t a b l i s h e d s o f t w a r e • . E a c h s y m b o l r e p r e s e n t s t h e m e a n ± s t a n d a r d d e v i a t i o n f o r 5 0 c o m e t s p e r s l i d e . T h e u n i t s a r e a r b i t r a r y a n d t h e s c a l e s h a v e b e e n s h i f t e d s o t h e v a l u e s a t t h e h i g h e s t d o s e a g r e e . 43 E s t a b l i s h e d So f tware New So f twa re CD 0 1 0 I ' I ' I ' I 1 I 1 I 1 I 1 I 1 I o 1 . 1 . 1 , 1 . 1 2 5 I 2 0 ^ 15 c E 10 0 * 5 1 0 1 • • • • 1 • * 1 11 • 1 1 11 O Q o o ,,,,9, o 4 0 3 5 \r £ 3 0 CD c 2 5 cu - i 2 0 •1' 1 5 10 5 0 o o o ,<9 o l o o n o o, o w @ o o o 'o o o o oc I I I 1 I I I o 0 1 2 3 4 5 6 7 8 9 150 100 r c 5 0 0 0 5 10 15 2 0 2 5 F l u o r e s c e n c e F l u o r e s c e n c e F igure 3 .17 C o m p a r i s o n of c o m e t m e a s u r e m e n t s us ing the new and e s t a b l i s h e d so f tware f o r an a s y n c h r o n o u s ce l l p o p u l a t i o n . E a c h s y m b o l r e p r e s e n t s m e a s u r e m e n t s of one c o m e t . The s a m e s l ide was m e a s u r e d s e p a r a t e l y f o r e a c h so f tware p r o g r a m . The ce l l s were not i r r a d i a t e d . 44 The new software indicates where the end of the comet tail was found during data collection, allowing the operator to compare the directly observed tail end position to the tail end detected by the new software. These were found to be in agreement; hence, it is certain that the new software makes measurements of the comet tail end as accurately as the operator. There is some discrepancy between the measurements of tail length and moment made by the new and established software. The curves of Figure 3.16 are smoother for the measurements made by the new software, especially for the tail length, for which the new software has been verified by direct observation as described above. Less variability is also seen in the measurements of asynchronous cells (Figure 3.17) with the new software, particularly for the measurements of S phase cells. Since the old software produces tail length values which are more widely varied, it is concluded that the new software is superior for detection of the comet tail. 45 4. Fluorescent Staining and Measurement Conditions 4.1 Introduction The DNA comprising the comet is detected by the fluorescence emitted by the bound stain. There were two reasons to investigate the characteristics of fluorescent staining and conditions of measurement: Changes in the staining conditions will affect the ability to detect the comet against the background and may improve assay sensitivity. In addition, the relationship between the amount of fluorescence and the amount of DNA present is not clear and is changed by the amount of DNA damage; understanding the mechanisms of staining may yield information on the state of the DNA in the gel. Fluorescent staining using intercalating (propidium iodide) and non-intercalating dye (Hoechst 33342), and fluorescent-labelled antibodies against incorporated bromodeoxyuridine have been found to give a comparable dose-response relationship in the comet assay (Olive et al., 1992). However, a decrease in comet fluorescence is observed for increasing DNA damage (Olive et al., 1994) indicating that some change in staining occurs with DNA damage perhaps due to DNA higher-order structure which has been shown to affect dye binding (Angerer & Moudrianakis, 1972). (While the comet assay utilizes propidium iodide (PI) for DNA staining, the characteristics of the propidium ion are very similar to the well-studied ethidium ion, so these are discussed interchangeably.) The ethidium ion binds to DNA in two modes: electrostatically bound to the negatively-charged phosphate groups, and intercalated into the DNA helix (pp. 350-365 in Saenger, 1984). The fluorescence emitted by the fluorophore depends strongly on the fluorophore binding mode, since the intensity of fluorescence is strongly dependent on the fluorophore environment. The quantum yield is low for the electrostatic binding mode, due to collisional quenching of the fluorophore by the solvent molecules; but while intercalated, the fluorophore is immersed in the hydrophobic region of the DNA and protected from this quenching; As well, energy for increased fluorescence may be 46 transferred from the DNA molecule to the fluorophore, a mechanism available only in the intercalated mode. There appears to be more than one mode of intercalative binding of PI to fixed, permeabilized cells with corresponding equilibrium constants K varying from approximately 3xl05 to lxlO7 M- 1 due to DNA conformation (Bertuzzi, et al. 1990). Under denaturing conditions in solution, the fluorescence is not increased above that of the dye, alone, in solution (pp.116-158 in Saenger, 1984; LePecq & Paoletti, 1967). This binding mode may, however, yield an increase in fluorescence under observation conditions of the comet assay while the fluorescent efficiency remains unchanged, since the clustering of fluorophore about the DNA molecule will cause an increase in the local dye concentration. This effect would be unseen in studies measuring average fluorescence of DNA in solution. The dependence of fluorescence on double-strandedness of the DNA has important implications for the alkali comet assay, where a loss of fluorescence is measured for comets from cells sustaining high DNA damage (Olive et al., 1994). This may mask the cell-cycle position as measured from total DNA content: highly damaged cells will have reduced total fluorescence, causing these cells to appear as cells earlier in the cell cycle. For example, highly damaged early S phase cells will appear to have the DNA content of undamaged Gl cells. This drop in fluorescence is believed to be due to incomplete renaturation of the DNA duplex, a conclusion supported by observations of DNA in solution, where calf thymus DNA denatured by heat (and renatured) shows a 50% drop in fluorescence (LePecq & Paoletti, 1967). The binding of ethidium and propidium occurs in a two-stage process. The kinetics of binding have been studied and are reported to involve a diffusion-controlled binding at the outside of the double-helix which precedes intercalation between the bases within the double-helix. Intercalation forces the neighbouring bases to unstack, distorting the sugar-phosphate backbone and altering the regular helix structure. With increasing amount of bound dye, the helix is lengthened and stiffened, a process which can be visualized by a 47 combination of pulling along the B-DNA double-helix and left-handed unwinding of 26 degrees per bound dye molecule, as necessary for the sugar-phosphate backbone to remain intact. Other factors influence binding of EB and PI to DNA, such as the presence of nucleoproteins, which mask binding sites (Angerer & Moudrianakis, 1972), and DNA structural changes (Prosperi et al., 1991). Based on this effect, PI has been used to probe DNA structure in intact cells (Bertuzzi et al., 1990). The interaction of these fluorophores with DNA is complex, however, since these molecules alter the coiling of the DNA duplex and influence its ability to renature and bind further dye (Bauer & Vinograd, 1968; pp.350-368 inSaenger, 1984). The double-strandedness of the DNA in the comet assay is dependent on the extent of renaturation occurring prior to or during staining, which depends on the extent of denaturation of the DNA due to previous alkaline treatment and the forces restoring the duplex. The DNA duplex is stabilized by a variety of interactions, which can be grouped into two classes: those interactions acting in the plane of the bases due to hydrogen bonding, and those due to interactions perpendicular to the plane of the bases, including hydrophobic interactions. Hydrophobic effects are due to base-stacking which minimize the exposure of the hydrophobic bases to the surrounding polar medium; these effects are reported to dominate the stability of the DNA duplex in polar solvent such as water, while hydrogen bonding plays the greater role in a nonpolar solvent (pp.116-159 in Saenger, 1984). The denaturation of the DNA duplex under extremes of pH are due to the protonation or deprotonation of the bases, which disrupts both hydrogen bonding and hydrophobic stabilization (pp.105-116 in Saenger, 1984). This disruption of the forces keeping the helix together causes the transition from helix to random coil structure. Under neutral conditions, the duplex has a greater charge density and hence a larger electrostatic potential; and under alkaline conditions, the coil configuration has a higher charge density than the helix (Bloomfield et al., 1974). Due to these differences in charge density for different pH, the effect of increasing ionic strength is to stabilize the duplex at neutral 48 conditions, and destabilize the helix at high pH. The G+C content also increases the helix stability since these bind more strongly than A+T regions. These regions denature at different temperatures, forming bubbles of denatured A+T regions held together by G+C regions not yet denatured. Since the pK values of the nucleotides are approximately <10.5 (extrapolated for zero ionic strength) (pp. 105-116 in Saenger, 1984), all hydrogen bonds for nucleotides are expected to be disrupted at the high ionic strength and pH used in the alkaline comet assay. This hypothesis was tested by observing comets stained under increasing pH conditions until denaturatioin occurred. After electrophoresis, the DNA is returned to neutral conditions where the single-strands will renature with the extent of renaturation depending on the ability of the strands to form hydrogen bonds with the matching complementary strands or strands of local complementarity. This renaturation occurs in two steps. The formation of a nucleus of three complementary base pairs must form first, this formation involving an unfavourable free-energy increase. The fourth base-pair to form is a favourable reaction, due to the stacking of bases, and further helix formation is rapid and involves negative free-energy changes (pp.116-159 in Saenger, 1984). Consequently, if the denaturation is incomplete, with small sequences of DNA duplex intact, the duplex will zipper together perfectly when returned to renaturing conditions. Otherwise, the two complementary strands are unlikely to rejoin completely. Small regions of paired complementary sequences will join which are not between proper complementary strands or which are between the complementary strands but out of register. These imperfect matches must dissociate before further renaturation can occur. By holding the DNA in solution at a temperature slightly below the melting temperature, imperfect matches are disrupted with only perfect matching strands remaining associated. Higher temperatures also allow faster diffusion and a greater likelihood of encountering the proper strand and region. During the alkaline comet assay, the strands also become separated and trapped by the agarose matrix, which can be expected to further limit renaturation. 49 The extent to which the complementary DNA single-strands will drift apart from each other during the comet assay will depend on several factors; the most important is the extent of unwinding of the DNA duplex. The strands of the DNA duplex cannot separate from each other without first unwinding, with this unwinding taking place at each single-strand break. Rates of DNA unwinding measured using hydroxylapatite chromatography indicate that unwinding is very slow in solutions of similar composition to the comet assay lysis and rinse. For unirradiated mammalian cellular DNA, only 5% is expected to unwind in 60 minutes in 0.03 M NaOH and 25% in 60 minutes in high salt solution (0.9 M NaCl) and 0.03 M NaOH (Rydberg, 1975). From these data and the model used for unwinding, DNA from cells irradiated with 0 or 1 Gy x-rays would require 6 hours or 24 hours respectively to unwind in solution. (See Appendix 4.1) Since the agarose gel matrix can be expected to hinder rather than assist in unwinding, these data suggest that unwinding largely does not occur in the alkaline comet assay at low levels of damage. Since the effect of high ionic strength is to increase the rate of unwinding nearly ten-fold (Rydberg, 1975), additional time spent in the low ionic strength rinse and electrophoresis will not yield much further unwinding than occurred during lysis. However, denaturation of the DNA duplex does not appear to require unwinding. Nuclear magnetic resonance (NMR) studies indicate that small regions of the stable DNA duplex spontaneously separate and swing out into the external medium, while the overall duplex remains wound (Adams et al., 1992). Considered with the heat-induced denaturation bubbles of A-T rich regions that exist within the duplex, this local denaturation indicates that unwinding is not necessary for denaturation. The final fraction of double-stranded DNA present in the alkali unwinding assay is therefore assumed not to be DNA that has resisted denaturation but rather DNA that has not separated in solution and renatures quickly. Therefore, while total denaturation is assumed to occur in the alkali comet assay, at low damage levels the strands will not unwind completely and are not expected to separate completely in the gel. 50 The effects of stain solution and dye concentration on comet measurements were investigated. Comet measurements made with PI were compared to those using the fluorescent dye, YOYO, which has been reported to give greater sensitivity than PI because of higher quantum efficiency (Singh et al., 1994). The binding of PI to DNA was modelled to estimate the number and proportion of fluorescence due to each binding modes. The effects of ionic strength, presence of EDTA and concentration of dye were varied to optimize measurement accuracy. Modification of the appearance of the comet by irradiation after electrophoresis was investigated as a means of improving camera utilization. 4.2 Methods and Materials The standard alkaline comet assay was used throughout except for the staining technique which varied as indicated below. Staining was performed with propidium iodide (PI) overnight (>8 hours) except for time-course experiments and for slides stained with YOYO (benzoxazolium-4-quinolinum, oxazole yellow homodimer). In experiments where slides were irradiated after electrophoresis, slides were immersed in water (neutral or alkaline) and irradiated using the Phillips 250 kVp x-ray machine at the B.C. Cancer Agency. The stability of comets was examined by measuring the same comet slides twice, 6 or 16 days apart (two sets of comparisons) for doses of 0,1 and 5 Gy,. The significance of the difference in means of these measurements was found using the t test. The error in point measurements was estimated from the coefficient of variation of 20 measurements per comet for four comets of different damage levels. For the effect of salt and EDTA in the staining solution, and for comparison of the fluorescent dye YOYO (0.6 ug/mL), multiple 1 Gy comet slides were prepared in a single batch of slides in one experiment. Slides were stained in 200 mL of solution, except for the slides stained with YOYO, which were stained by placing 0.5 mL of stain (2.5 ug/mL) on each slide and allowing the slide to partially dry. Assuming a gel volume of 1.5 mL, this 51 gives a concentration of 0.5 u.g/mL. For the staining time-course, single slides were removed from the staining solution at the appropriate time and placed in storage boxes to be observed later. Two experiments examined the effect of stain concentration on comet values, using sets of six to ten slides for doses of 0, 1 and 5 Gy irradiation. The cell suspensions were irradiated at high concentration, then diluted and split into 0.5 mL samples. These slides were stained overnight (>8 hours) in 0.02 M NaCl with PI concentration from 0.025 to 10 iig/mL. Following staining the slides were placed in cold storage and observed over the next two days except for the second experiment where the slides were observed immediately after removal from the staining solution. The order of slide preparation was randomized (out of order of dye concentration) to ensure that order effects did not dominate the results. The first experiment (Set 1) was performed before, confounding factors (such as apparent increases in damage due to extended alkaline rinse and cell suspension) were examined. The timing of slide treatment in each step of the assay was most accurate in the second experiment (Set 2). The effect of staining conditions on comet measurements was determined by direct measurement and by comparing the expected influence according to the image analysis. The software distinguishes the comet from the background using the comparison of the comet element fluorescence, At, to the threshold (defined as the background mean plus three standard deviations), T. The comet element is detected if At > Tor (At,/ T) >1. The greater the (At - 7) or {At / 7), the greater the ability to distinguish the comet element from background. Note that (Aj-T) depends on the camera gain, a higher gain yielding a larger difference. With At and T expected to depend mostly on the fluorescent image, an increase in gain will both increase the comet element value and background to the same extent; therefore, this ratio is expected to be largely independent of gain setting. Differences in comet measurements such as area or tail length will indicate differences in comet detection. 52 From elementary chemistry, for a macromolecule with independent binding sites, the equilibrium relation between an empty binding site S, a dye molecule free in solution D and a site-dye complex C is S + D++C (1) with equilibrium binding constant K, K=[C]/([S\[D]) (2) where square brackets denote concentration. To place the relation on a per-DNA-phosphate basis, the total DNA phosphate of the molecule is assigned P. Defining, r=[C]IP, n = [C+S]/P, and c=[D] (3) the value K can be re-written K = rP/(n-r)Pc (4) oxr/c=Kn-Kr (5) which is the well-known Scatchard relation. Since dye is present in great excess compared to the amount of DNA, c is taken as both the initial concentration and the free-dye concentration after equilibrium binding has occurred. The amount of DNA in a normal mammalian cell in culture is assumed to be constant, so the value P is constant for all cells and comets for a cell sample of plateau-phase cells. The fraction of available binding sites which are bound by dye from (5) is r/n = cK/(\+cK) (6). An absolute measurement of r cannot be found with the equipment used but assuming that the fluorescence, F, is proportional to the amount of bound dye, F=kr. The relationship between the measured fluorescence and the dye-binding parameters is: F = kncK/(\+cK) (7). Since k is unknown, the absolute number of binding sites, n, cannot be determined. The concentration of PI was calculated based on a molecular weight of 668 g/mole (from Molecular Probes Handbook of Fluorescent Probes and Research Chemicals by Richard P. Haugland 1992-94) by multiplying the mass per volume in u.g/mL by 1.50xl0"6. Comet 53 fluorescence was modelled according to the above relation for one or two independent binding modes. Modelling of data was performed using the Levenberg-Marquardt method for nonlinear modelling as described by Press et al. (pp. 683-688, 1992) utilizing their computer software routines. The chi-squared goodness of fit parameter and the probability of this large a chi-squared occurring by chance if the model is correct (the Q value) are reported (Q values close to 1 indicate a good fit; Q values close to zero indicate an unsatisfactory fit.) The effect of stain pH was investigated by titrating 1 M NaOH into a 40 mL solution of 2.5 ug/mL PI of originally 0.02 M NaCl or 1 M NaCl. A single comet was observed under the fluorescent microscope for each profile. pH values were measured with test strips after addition of 0.1 to 0.5 mL NaOH using an Eppendorf micropippette. The camera gain setting was set to the maximum without saturation at the start of each profile. Slides were stained overnight before addition of NaOH. After each addition of 0.1 to 0.5 mL of 1 M NaOH, the pH was measured and several comet measurements were made over a few minutes as the solution and gel came to equilibrium. 54 4.3 Results 4.3.1 Comet Stability and Measurement Error Slides of comets for 0, 1 and 5 Gy of damage were measured several days apart. The significance of the difference of means of comet measurements was determined using the t test (Table 4.1). Only one slide showed a significant difference in tail length (at significance 0.005) but for the same set of data, the tail moments were not significantly different (at significance 0.99) for this slide (1 Gy, 6 days apart). The reason for this discrepancy is not known. The comet fluorescence was significantly different for all slides. These data suggest that comets are generally stable for periods of time of weeks except for comet fluorescence, which may not be stable due to exposure to distilled water in storage or changes in camera response. The point measurement error was estimated by repeatedly measuring the same comet 20 times. The coefficient of variation (the standard deviation divided by the mean) for each set of data is shown in Table 4.2. Measurements were made for four comets of different damage levels judged by direct observation using the microscope ocular. The "faint tail comet" had a tail only slightly visible when viewed directly. The comet next higher in damage had a faint tail, which was clearly seen when viewed directly. The 5 Gy comet had a very distinct, bright tail. During data collection, the end of the comet tail identified by the software was in good agreement with the tail end estimated from direct observation. 55 Table 4.1 Comet Stability with Storage * Dose Days Between Tail Moment Tail Length Fluorescence (Gy) Measurements 0 16 0.11 0.56 O.01 1 16 0.76 0.43 O.01 1 6 0.99 0.005 O.01 5 6 0.31 0.61 O.01 * Significance ol the difference ot means (t test) tor the same comet slide measured several days apart. Table 4.2 Point Measurement Error for Comets of Different Levels of Damage * Damage Level: No Visible Slight Visible Faint Pronounced Tail Tail Tail Tail (5Gy) Tail moment 0.085 0.12 0.19 0.11 Tail length 0.46 0.11 0.14 0.053 Comet fluorescence 0.064 0.034 0.049 0.0090 * Values represent the coefficient of variation (standard deviation/mean) tor 20 measurements made on the same comet. Four comets were measured. 4.3.2 Staining Time and Ionic Strength of Solution Staining of the DNA is dependent on the time spent in the dye solution, the ionic strength of the solution, and the dye concentration. The staining had not reached steady-state after 2 hr when stained in 2.5 ug/mL PI at low ionic strength. The mean and variability of the background increases dramatically with increasing duration of staining (Figure 4.1). The comet fluorescence is seen to decline with increasing duration of staining (Figure 4.2). It is assumed that the increase in fluorescent background is due to the increase with time of the dye present in the gel due to increased amounts of free dye in the gel or associated with the gel. The decrease in comet fluorescence with increasing staining time occurs simultaneously with increasing background, suggesting that the decline in comet 56 O C <D O CO <D l _ o CO 3 •*-> iZ c 3 TJ a >N ino JDJ i _ -t-> kg rbi o o o op c D 0) 2 1 r J2 o c 2.0 3 c ° 8 5 1.6 co J-Q) o Is" ^ *-M 1.2 -o .2 c > 3 <l> 1.0 o Q v i_ O.o o o O X J OQ c 0.6 D (7) 0.4 T 0 M NaCl • 0.02 M NaCl O a) / i , , , . i . 0 M NaCl • 0.02 M NaCl O b) 0 5 20 25 Duration of Staining (hr) 30 Figure 4.1 The effect of staining duration on background screen f l u o r e s c e n c e , a) The mean of the background f luorescence , b) the variation in the background f luorescence measured by the s tandard deviation of the background for each image. The symbols indicate the mean ± standard deviation of 50 measurements of 1 Gy c o m e t s . Stain concentrat ion was 2.5 y^g/mL, in either 0 or 0.02 M NaCl . 57 1.1 CD O c CD O CO CD o 1.0 Z5 -+-' CD •E o o CD >^ 0.9 D CD cn 0.8 i i | , — 0 M NaCl • 0.02 M NaCl O 0 i i 1 1 1 1 i -7 / 0 5 20 2 5 30 Time of S t a i n i n g ( h o u r s ) Figure 4.2 The effect of staining duration on comet f luorescence. The comet f luorescence for dif fering durat ions of staining are shown normal ized to the f irst measured values. Each symbol represents the mean ± SEM of 50 comets per s l ide. Comets were stained 2.5 jug/mL PI in 0 or 0.02 M NaCl. 58 fluorescence is due to the absorption by free or agarose-bound dye of fluorescence emitted from the comet DNA during observation. When 0.02 M NaCl is included in the staining solution, the maximum background is reduced to only 20% of the maximum background without NaCl. Presumably these positive ions compete for binding to negatively-charged chemical groups which are constituents of agarose gel (Kirkpatrick, 1990). An increase in comet fluorescence occurred on addition of 0.02 M NaCl, which is contrary to expectation; positive ions such as Na+ are known to suppress the electrostatic binding of PI to DNA which precedes the highly fluorescent intercalative binding. Increasing the concentration of NaCl to 0.1 M NaCl suppresses the comet fluorescence, as expected. The presence of EDTA, which chelates heavy metal ions that compete with propidium/ethidium for nucleic acid binding, has no measurable effect at low concentrations while at high concentration tends to reduce the comet fluorescence (Figure 4.3). 4.3.3 Stain Concentration The effect of PI concentration on comet measurements was measured for 0, 1 and 5 Gy comets. Tail moment shows a slight increase and tail length is constant for increasing PI concentration (Figures 4.4 and 4.5) while total area decreases slightly and tail area is constant (Figure 4.6). Intercalating dye such as EB and PI are known to increase the extent of renaturation of DNA (LePecq & Paoletti, 1967). This phenomenon seems to play a role in retraction of tails seen under other comet assay conditions but is not visible here. The earlier comet assay performed electrophoresis under non-denaturing conditions (pH 8.3) (Olive et al. 1990). It is possible that the previously seen tail retraction was due to a greater extent of double-stranded DNA, while alkaline lysis and alkaline electrophoresis may prevent this effect. Comet fluorescence shows an increase for increasing concentration of PI (Figure 4.7). The threshold depends on both the mean and standard deviation of the fluorescent background and shows an increase with PI concentration (Figure 4.8). The ability to 59 ce 2 5 c / N CD CO j J O CO c CD 3 2 0 o >v Z3 1— EE D i_ - M -t-> CD ' -£15 E o o o 10 6 "D 1 5 C c no 3 4 CD O 3 o D m ->r-> *-Q 2 D 1 0 o o o o o o o o o o F igu re 4 . 3 The e f f ec t of NaCl and EDTA on c o m e t s t a i n i n g . The e f f e c t of EDTA and NaC l in the s t a i n i n g so l u t i on of 2 . 5 ^ a g / m L on (a) c o m e t f l u o r e s c e n c e and (b) b a c k g r o u n d a re s h o w n . The b a r s i n d i c a t e the m e a n ± S E M of 5 0 c o m e t s per s l i d e . C o m e t s were s t a i n e d ove rn igh t (>8 h o u r s ) . 60 PI concentration (/^g /rr iL) Figure 4.4 The effect of PI concentrat ion on tail moment measurements . Sl ides were stained for >8 hours In 0.02 M NaCl at the indicated concentrat ion of PI. Symbols represent the mean ± SEM of 50 comets per s l ide. 61 =1 180 160 140 120 100 cn c 80 CD 60 40 20 0 - i i i I - 0 Gy data • _ - .1 Gy data V -- 5 Gy data . T -- T T - • T • • -- • -- V -- -- V V -V V V -- : * i " --- i i : -• I I i i I i 0 8 10 12 PI concentrat ion ( j L t g / m L ) Figure 4.5 The effect of PI concentrat ion on tail length measurements . Slides were stained for >8 hours in 0.02 M NaCl at the indicated concent ra t ion. Symbols represent the mean ± SEM of 50 comets per s l ide. 62 CN E =1 D CD < D O CN =1 D CD < 11000 10000 9 0 0 0 8 0 0 0 7 0 0 0 6000 5 0 0 0 4 0 0 0 3 0 0 0 1 1000 10000 9 0 0 0 8 0 0 0 7 0 0 0 6000 5 0 0 0 4 0 0 0 3 0 0 0 2 0 0 0 1000 0 0 Gy • 1 Gy v 5 Gy • 0 2 4 6 8 10 PI concen t ra t ion ( j i t g / m L ) Figure 4.6 The effect of PI concentrat ion on area measurements , a) Total area of comet image, b) Tail area of comet image. Symbols indicate the mean ± SEM of 50 comets per sl ide. 63 2 5 co o c CD _ <-> CO CO -*-> £ *E O =5 -2 >, CD ° D 2 0 15 0 Gy da ta 1 Gy da ta 5 Gy da ta v 10 0 10 PI concentrat ion ( / i g /mL) Figure 4.7 The effect of PI concentrat ion on comet f luorescence measurements . Sl ides were stained for >8 hours in 0.02 M NaCl at the indicated PI c o n -cent ra t ion. Symbols represent the mean ± SEM for 50 comets per slide of radiation doses of 0, 1 and 5 Gy. 64 8 c zs >s 6 D ^ 5 o 0 Gy data 1 Gy data 5 Gy data v v f V O sz CO CD • V - 2 0 8 10 12 PI concen t ra t ion ( y U g / m L ) Figure 4.8 The image threshold variat ion with PI concent ra t ion. The threshold is equal to the mean plus three standard deviations of the background f luorescence, measured at the top and bottom rows of the image. Each symbol is the average of values for 50 comets per sl ide. Comets were stained overnight. 65 distinguish the comet from the background is dependent on the image threshold, itself a function of the background mean and variability. The average comet element fluorescence (the comet fluorescence divided by the area in elements) compared to the threshold (mean plus three standard deviations of the background) shows an increase with PI concentration (Figure 4.9). The difference between the average comet element fluorescence and the threshold indicates the ability to resolve the comet image. However, the calculation of the average comet element fluorescence is based on calculation of the comet area, itself determined from the threshold. Hence, a high threshold can be expected to reduce the calculated area, leading to a increased average comet element fluorescence, and a inappropriately high calculated value for the average comet element fluorescence above threshold. The slight decrease in area for increasing PI concentration (Figure 4.6) is consistent with the increase in threshold (Figure 4.8) suggesting that the slight increase in average comet element fluorescence above threshold for increasing PI is an artifact. A different measure of the ability to distinguish the comet image from background is the ratio of average comet fluorescence to the threshold, a measure that is independent of gain setting. This measure shows a clear increase with decreasing PI concentration in Figure 4.10a). A similar measure that is independent of the comet area is the ratio of expected comet fluorescence to the threshold; this ratio also increases at low PI concentration (Figure 4.10b). (The expected comet fluorescence is calculated independently from the comet area and threshold using the entire screen fluorescence and subtracts the background, unlike the usual comet fluorescence.) The binding of dye to DNA was modelled using one or two binding modes, with an equilibrium constant and a scaling factor for each mode. Comparisons of these models to the data are shown in Figure 4.11. These data are from a second experiment which was performed after some confounding factors were eliminated (described in Chapter 5). The data for the first experiment did not give a reasonable fit to the model due to large scatter, possibly due to the storage of slides before measurements were made. As indicated in 66 40 co 35 -+-> c CD O c ^ 3 0 CD O o L_ CO •+-> CD • — 0 P - £ 25 L Z XI CD o 2 0 E sh o CD o ^ • 1 5 CD 1— cn ve o ve CD o 1 0 > _Q < O 0 Gy data 1 Gy data 5 Gy data V V V V v V 0 8 1 0 PI concen t ra t ion ( / ^g /mL) Figure 4.9 The effect of PI concentrat ion on the di f ference between the average comet element f luorescence and the threshold. The threshold value is the background mean plus three standard deviat ions. Symbols are the mean ± SEM values for 50 comets for each sl ide. 67 PI concentrat ion ( j i ig/mL) Figure 4.10 The effect of PI concentrat ion on comet f luorescence to threshold ratio. The comet f luorescence is the image element average value. The expected comet f luorescence is independent of the comet area (background subt racted) . Symbols represent the mean of 50 comets per s l ide. 68 0.0 5.0 10.0 15.0 20.0 0 PI concentrat ion (10 M) Figure 4.11 Comet f luorescence for varying dye concent ra t ion. PI concentrat ions are f rom 0.025 to 10 /i-g/mL converted to M. The symbols are means ± SEM for 50 comets . The data were fit to models with 1 or 2 independent binding modes. 69 Table 4.3, these data from the first experiment show a poor fit to the model as indicated by high chi-squared values and low Q values (which is the probability that this large a chi-squared would occur if the model is correct). The value of Q is not directly interpretable here, since it assumes both normal distributions of measurement errors in comet fluorescence and an adequate representation of these errors. The measurement errors of the comet fluorescence were the standard error in the mean (SEM) of the measured values which are not expected to follow a truly normal distribution; more significantly, additional measurement error that is not represented in the SEM is expected to be present due to changes in camera response between observation of comet slides. Hence, the values of chi-squared and Q are thought to be overly pessimistic. The single binding-mode model is clearly insufficient to explain these data in Set 3 since the curves do not pass through the data in Figure 4.11. The model with two modes of binding fits the data well. Approximately 20% of the fluorescence is due to the lower affinity binding mode as indicated by the scaling factors, which represent the products of the number of binding sites and the proportionality between comet fluorescence and amount of bound dye. (Hence, the number of fluorescent binding sites cannot be determined independently of the fluorescent yield of that binding mode.) In both data sets a decrease in the number of high affinity binding sites occurs with increasing radiation dose. The values for the equilibrium constants found for Set 2 for two binding modes (3.7 ± 0.2 xlO5 and 3.6 ± 0.2 xlO7) are in general agreement with values reported for binding sites in permeabilized whole cells of 3.0 ± 0.2 xlO5 M- 1 and 1.1 ± 0.4 xlO7 M- 1 (Bertuzzi etal., 1990). 70 Table 4.3 DNA-Dye Binding Parameters Dose Scaling Factor Equilibrium Constant Goodness of Fit Q a (Gy) (xl07M-i) (chi-squared) Set 1 0 1 5 21.85±0.07 20.48±0.06 17.00±0.07 4.0±0.5 2.8±0.2 1.19±0.07 170 240 670 6x10-32 0 0 Set 2b 0 1 5 24.57±0.08 23.00±0.06 20.28±0.06 3.33±0.09 2.73±0.07 2.25±0.05 206 162 542 1x10-42 6x10-34 0 Set 2b 0 (1) 22.7±0.2 4.2±0.2 (2) 4.5±0.5 0.020±0.0'08 0.62 0.99 1 (1) 20.6±0.7 3.3±0.3 (2) 3.5±0.6 0.072±0.035 24 9x10-5 5 (1) 17.7±0.2 3.3±0.1 (2) 5.7±0.3 0.019±0.004 38 5x10-7 a Q is the probability oi having this large a chi-squared and a correct model, assuming normal distributions of errors. Low values indicate the model is not supported, values close to one indicate a good fit to the model. b These data were taken from an experiment after sources of variation had been isolated. These data are fit to one or two binding modes. The separate binding modes are indicated by (1) and (2). 4.3.4 pH of Staining Solution The expected comet fluorescence (the total fluorescence with background subtracted) of a 5 Gy comet was measured for increasing pH at a PI concentration of 2.5 ug/mL at high and low ionic strength (Figure 4.12). The fluorescence values that lie vertically represent measurements made over a period of several minutes after addition of NaOH solution. The comet fluorescence, which has been normalized by the pH 7 value, shows a strong drop in fluorescence at the region where DNA denaturation occurs. The presence of PI will stabilize the helix and shift the alkaline denaturation points to higher 71 1.2 ~r——i r CO 1 ' ° o c cu o OT 0.8 CD O o Z5 CD E o o <D > D 0.6 0.4 a) 0.2 LY. o o 0.02 M NaCl O 1 M NaCl • 83 O o , ° o 0.0 _ l I L _ _l i l _ 8 10 12 14 pH of Stain Solut ion Figure 4.12 The effect of pH of stain on comet f luorescence. A single 5 Gy comet was measured repeat edly for each profile (0.02 or 1.0 M NaCl stain solut ion) while 1'M NaOH was added. Several measurements were taken after NaOH was added. Lower points at the same pH were taken at later t imes as the gel pH changed. 72 values than in solution without PI (LePecq & Paoletti, 1967). The lowest fluorescence point of the 0.02 M NaCl profile represents the residual fluorescence after the comet image has disappeared from view, likely due to uneven background fluorescence. Since the comet became invisible at the highest pH and a two-binding mode model was sufficient to describe the data, electrostatically-bound dye apparently contributes little to the comet fluorescence. The denaturation of the DNA duplex is assumed to occur at the steepest decline in the profile. 4.3.5 Comparison of PI to YOYO The comet fluorescence and background fluorescence of slides stained using YOYO or PI were measured. Slides for each of the two experiments (Tables 4.4 and 4.5 respectively) were prepared from one cell population. While YOYO produces comets of higher fluorescence and lower background mean and variation across the screen, the variation in comet fluorescence is much larger for YOYO compared to PI staining at low ionic strength. The CV of comet values are similar or improved for YOYO (Table 4.4 and 4.5) compared to staining with PI with 0.02 M NaCl. But the coefficient of variation of tail moment between 0 and 1 Gy measurements (the sum of the standard deviations at 0 and 1 Gy divided by the difference in means) for YOYO is larger (1.75) than for PI at 2.5 ug/mL (1.14). This indicates that YOYO staining does not result in greater sensitivity, a result contrary to the results reported by Singh et al. (1994). 73 Table 4.4 Variability of Comet Measurements Due to Staining (Coefficient of Variationc) Dose StainVconcentration Gain Tail Tail Tail Total (Gy) (ug/mL) Settingd Moment Length Area Fluorescence 1 PI / 0.25 low 0.46 0.11 0.31 0.05 1 PI / 0.25 high 0.35 0.08 0.21 0.04 1 PI / 2.5 low 0.37 0.10 0.30 0.04 1 PI" / 2.5 low 0.40 0.06 0.20 0.09 1 PI / 10 low 0.41 0.15 0.30 0.03 1 YOYO /0.5 low 0.42 0.07 0.16 0.07 5 PI / 2.5 low 0.13 0.07 0.08 0.08 5 PI / 2.5 high 0.11 0.06 0.05 0.03 a Stain solutions or PI are 0.02 M NaCl except where noted, b Stain solutions of PI are without M NaCl. c coefficient of variation of comet value = (standard deviation)/ mean for 50 comets per slide. d Gain setting of low corresponds to a gain setting where the 0 Gy slide of this stain treatment is maximized without saturating the camera; high gain setting refers to a gain maximized for the individual slide. Table 4.5 Comparison of Staining with YOYO and PI Dose CV* Ratio of Comet Fluorescenceb (Gy) to Threshold PI YOYO PI YOYO 0 0.56 0.59 4.47 9.95 0.5 0.48 0.40 3.97 8.10 1 0.40 0.42 3.61 7.84 a CV is the coefficient of variation (standard deviation/mean). b Threshold is the mean plus three standard deviations of the background. 4.3.6 Effect of Camera Settings The ability to distinguish the comet from background is also influenced by the camera gain setting. The use of low dye concentrations enables the camera gain to be 74 increased. Setting the gain to the maximum without saturation of the camera is expected to maximize the ability to distinguish the comet; therefore, measurements of dim comets are expected to be improved by using an optimum gain setting. Comets were examined which were dim due to the fluorescence being spread through the tail in a 5 Gy comet slide and due to low concentration of dye (0.25 ug/mL PI in 0.02 M NaCl) for 1 Gy comet slide. Increasing the camera gain leaves the ratio of average comet element fluorescence to threshold unchanged (difference in means not significant at P>0.10), but increases the difference (significant at PO.001), as seen in Figure 4.13. The increase in gain appears to increase the ability to distinguish the comet from background, yielding increased comet measurements as shown in Figure 4.14. This increase in area and tail length due to gain increase suggests that the comets were more accurately detected at a higher gain. 4.3.7 Effect of Irradiation Before Staining Irradiation of slides after electrophoresis (prior to staining) was performed as a possible means of increasing detection of the tail region by spreading the intensely bright head region to a larger area. This allows the camera gain to be increased with the benefit for detection of the tail as for measurement of dim comets described above. Radiation applied to comet slides after electrophoresis seemed to cause the head DNA to relax and expand. Figure 4.15 shows the effect on comet measurements of irradiation of slides prior to staining, after electrophoresis. Slides were irradiated with 10 Gy of x-rays after being placed in pure water or rinse solution for 1 minute. Irradiation under neutral conditions failed to decrease comet fluorescence significantly; while under alkaline conditions, the fluorescence is substantially reduced. The halo width and total area are dramatically increased for irradiation under both conditions, suggesting that the head DNA is held in a compressed state; after irradiation it is able to relax and spread into the surrounding agarose. The tail length is reduced by irradiation before staining under both neutral and alkaline conditions. This reduction in tail length is likely due to poor detection 75 Figure 4.13 The effect of gain setting on dim comet signal compared to threshold. Dim comets at high dose (5 Gy, 2.5 /xg /mL) and low stain concentrat ion (1 Gy, 0.25 yag/mL) were observed at two gain sett ings. Low gain sett ing was the maximum for a 0 Gy sl ide; high gain was the maximum for the current sl ide. Error bars indicate the error propagated from the standard errors of the average f luorescence and threshold for 50 comets per symbol . The dif ferences due to the gain sett ing are s i g n i f i -cant at <0.001 for (signal — threshold) and not s igni f icant at >0.1 for (s igna l / th resho ld) based on t test for the di f ference of means Stain was PI in 0.02 M NaCl. (Comet s ignal is the average of the comet element f luorescence.) 76 9000 Dose: 5 Gy 5 Gy 1 Gy 1 Gy Dose: 5 Gy 5 Gy 1 Gy 1 Gy Gain: low high low high Gain: low high low high Figure 4.14 The effect of gain sett ing on measurement of dim comets . 5 Gy comets were dim due to large area (stain is 2.5 / xg /mL PI). 1 Gy comets were dim due to dilute stain (0.25 fig/mL PI). Low gain was the maximum for 0 Gy, 2.5 / ^g /mL PI. High gain was the maximum for the individual s l ide. Symbols indicate the mean ± SEM for 50 comets per sl ide. 77 Figure 4 .15 The effect of Irradiation before staining on measurements of asynchronous V79 cel ls . Cell suspensions were Irradiated with 0 or 1 Gy, then with 0 or 10 Gy x - r a y s in alkaline or neutral solut ion after e lect rophoresis . The symbols indicate the mean ± SEM for 150 comets per s l ide. 78 of the more diffuse tail end after irradiation. This suggests that tail retraction does not occur in the alkaline comet assay, since irradiation before staining should separate the tail from the head and prevent retraction, yielding an increase in tail length. This is contrary to the tail retraction that is observed for the neutral comet assay (Olive and Banath, 1995). Irradiation with 1 Gy of x-rays followed by several minutes in alkali also results in spreading of the head region into a larger area. Irradiation of the slide prior to staining resulted in a slight increase in tail moment compared to the standard method (Figure 4.16), an effect likely due to the decrease in head fluorescence. (This irradiation lowers the total comet fluorescence, which is used to normalize the tail moment). The measurement of tail length is not enhanced. Total area is dramatically increased, apparently due to DNA expanding out from the head, since tail area is nearly the same. Both the background fluorescence mean and variation are increased independent of the radiation dose to the cell suspensions, resulting in a dramatically increased threshold for the irradiated slides (mean threshold ± SD for 4 slides of the dose-response curve: 5.00 ± 0.12 and 2.92 ± 0.12 for the irradiated and unirradiated slides respectively). The increase in threshold and decrease in fluorescence indicates a detrimental change for comet measurement as indicated by the signal to threshold difference and ratio (Figure 4.17). Asynchronous V79 cells were irradiated before staining with 1 Gy x-rays followed by 5 minutes in alkali. Figure 4.18 shows the effect of irradiation prior to staining on bivariate plots of tail moment and fluorescence for several levels of damage. (Note that the measurements were performed on a different gain setting for the slides which were irradiated or unirradiated before staining.) Approximately the same decrease in comet fluorescence occurs between the 0 and 0.25 Gy comets for irradiated and unirradiated slides (approximately 10% decrease for both). 79 pre-stain Irradiation standard assay O c CD E o C CD 0.00 0.25 0.50 0.75 1.00 Dose (Gy) 0.00 0.25 0.50 0.75 1.00 Dose (Gy) Figure 4.16 The dose —response for sl ides i r radiated before sta in ing. Sl ides were Irradiated with 1 Gy or 0 Gy (standard assay) . Standard assay sl ides were measured at a lower gain. Symbols are the mean ± SEM for 50 comets per sl ide. 80 Dose (Gy) Figure 4.17 The effect of irradiat ion of sl ides before staining on the signal to threshold ratio and di f ference. Slides were irradiated with 0 or 1 Gy in alkal i . Control sl ides were measured using a lower gain set t ing. The signal is the average comet element f luorescence. Each symbol is the mean of 50 comets . The threshold is the background mean plus three standard deviat ions. 81 Standard Assay Irradiated before staining Fluorescence (arbi t rary units) F luorescence (arbi t rary units) Figure 4.18 Modif icat ion of the cell —cycle measurement by Irradiation before staining. Asynchronous V79 cel ls were i rradiated with 0 Gy (standard assay) or 1 Gy x —rays in alkal i . The dotted lines are to aid the eye. 100 comets are shown per sl ide. 82 4.4 Discussion These results suggest that the ability to distinguish the comet image from the background is enhanced by reducing dye concentration, including 0.02 M NaCl (to reduce binding of the dye to the gel), and by optimizing the camera gain setting for each microscope slide if the population of comets are sufficiently homogeneous. Propidium iodide appears to give better sensitivity than YOYO due to lower variance of comet measurements. Treatment of comet slides with DNA-damaging agents immediately before staining decreases the intensity of comet head fluorescence, allowing the use of increased camera gain. The increased background fluorescence, however, detracts from any improvements in the ability to distinguish the comet from the background obtained from the higher gain. There appears to be two modes of binding of PI to DNA in the comet assay. The equilibrium constants correspond in general to the two intercalation binding modes described by Bertuzzi et al. (1990) due to binding to DNA in solution (low affinity) and to DNA bound to protein (high affinity). Under the conditions of the comet assay, chromatin is not thought to persist; and the high-affinity site may be due to a DNA conformation in the gel similar to that in chromatin. The equilibrium constant of the low affinity binding mode corresponds well to the accepted value for DNA in high ionic- strength solution. The reduced fluorescence due to the high affinity binding sites appears to explain the loss of comet fluorescence for increasing DNA damage. From the dependence of the comet fluorescence on the stain solution pH, denaturation of the DNA helix appears to occur at pH 11 to 12. Electrostatically bound PI appears to contribute little to comet fluorescence, since a two-mode model describes the data well and the comet disappears after denaturation. The concentration of NaCl (0.02 M) is expected to strongly inhibit this effect since only 3.8 uM PI is present (at 2.5 ug/mL). The time spent under alkaline conditions appears to be inadequate for complete DNA unwinding in solution ((Rydberg, 1975) and see Appendix 4-1). The constraints 83 placed on the DNA strands by the gel are expected to further retard unwinding. The expansion of the comet head following irradiation (after electrophoresis) suggests a highly constrained state, not conducive to unwinding. This lack of unwinding after complete denaturation suggests that the complementary DNA strands cannot separate completely during electrophoresis, though they may be sufficiently moved out of register in base-pair alignment for renaturation to occur only between random short stretches of complementary bases. 84 Appendix 4.1 - Rate of Alkali Denaturation of DNA According to Rydberg (Rydberg, 1975), the relationship between the fraction of DNA, F, found to be double-stranded during hydroxylapitite chromatography after allowing for unwinding in alkali for time t is related to the molecule weight, M, by ln F = - (constant) t0-66 / M. Using Rydberg's data for DNA in 0.03 M NaOH, 0.9 M NaCl (Rydberg, 1975) for DNA from unirradiated mammalian cells, F = 0.75 at t = 60 minutes, yielding the value for (constant) / M- 0.0193. From this data, 10% of the DNA is expected to be double-stranded after 24 hours in alkaline solutions with high salt, similar to the comet assay lysis solution (for F=0.1, t = 1400 minutes). For cells irradiated with 1 Gy x-rays, (constant) / M= 0.0465 and 6 hours are required for 90% of the DNA to unwind in solution, based on these data. 85 5. Effects of Assay Technique on Measurement of Induced Damage 5.1 Introduction In a typical experiment, the effect of a DNA damaging agent is measured for a range of doses of an agent. This requires the preparation of a number of samples within the same experiment. Obviously, differences in preparation and timing of the assay protocol between samples must occur and may affect the subsequent measurements of damage. The effect of these small differences in treatment of samples was investigated to determine their relative contributions to differences in DNA damage measurements which would limit the detection of low levels of DNA damage from ionizing radiation. In the assay, sets of cell samples are electrophoresed simultaneously but must be prepared sequentially; hence, each of the slides is treated slightly differently and one of the assay steps must be of a different duration for each sample. The purpose of the experiments described below was to identify those steps in the assay protocol which most strongly affect the detection of the DNA damage of interest. The comet assay measurements were the only means used to measure DNA damage. Differences in comet measurements could be due to actual differences in DNA damage induced by the assay itself, to differences in the assay response for the same levels of DNA damage, or a mix of the two. The changes in comet measurements due to changes in assay protocol are referred to as "apparent damage" since they produce changes in comet measurements which normally correspond to different levels of DNA damage. The assay protocol varied slightly between experiments as modifications were made in attempts to improve sensitivity and reduce variability. Apparent damage from the following sources were examined: the duration of cell-suspension, lysis and rinse; the type of cell suspension medium; and the agarose gel temperature and gel preparation technique. 86 5.2 Methods and Materials Four independent experiments were performed to measure the effect of storage of cell samples for different durations in different suspension media. These experiments were performed using slightly different protocols but each allowed comparison of effects of cell suspension within each experiment. One experiment used cells collected with trypsin, suspended in PBS and irradiated with 0.5 Gy using the Cs 1 3 7 unit. Cells were then transferred to PBS or growth medium (MEM + 10% FCS), mixed with agarose and transferred to lysis solution or held in suspension on ice for 2 hours either under fluorescent light on a lab bench or in darkness. The rinse duration was 1 hour. A second set of cells was collected with trypsin, suspended in growth medium (MEM + 10% FCS), diluted in 10 mL of MEM, growth medium or PBS. These were divided into 5 mL tubes (0.5 mL each), irradiated, then held on ice for 0, 45, 90 or 135 minutes under fluorescent light or darkness. Alkaline rinse duration was 1 hour. For a third experiment cells were collected with trypsin, suspended in growth medium at high density, diluted in 10 mL of cold PBS or growth medium. These were split into four small tubes for 0 or 1 Gy irradiations on the Csl37unit. The alkaline rinse duration was 3 hours. The agarose gel preparation technique seemed to have a strong effect on apparent damage in preliminary experiments. To examine this effect, cell samples were prepared and the agarose-cell suspension was allowed to cool for various times during mixing for two temperatures of stock agarose. Agarose (1 % w/v agarose in PBS) was prepared at 40 and 55°C. Unirradiated samples were rapidly injected into the agarose and allowed to cool in the lab room temperature air for up to 20 seconds before being pipetted up and down twice and poured onto a microscope slide. This experiment used an alkaline rinse of 3 hours. The effects of duration of lysis and alkaline rinse were investigated by comparing comet measurements produced by the standard assay protocol but using lysis and alkaline rinse durations from 0.5 to 3 hours. Three independent experiments were performed, with radiation doses of 0 and 0.25 Gy irradiated on the Cs 1 3 7 unit. 87 5.3 Results 5.3.1 Effects of Cell Suspension and Light Exposure The cells suspended in PBS show large amounts of apparent damage. The results of three independent experiments (Figure 5.1a) and b)) for cells suspended in PBS show an increased tail moment compared to the cells suspended in MEM and MEM+10% FBS (growth medium) for each experiment. The differences in tail moment between the cells suspended in PBS and growth medium are shown in Figure 5.1b). The rate of increase of tail moment for cells suspended in PBS compared to cells suspended in culture medium was 8.9 ± 2.5 |i.m/hour from linear regression; the slope is statistically significant (significance ofFis 0.016). 5.3.2 Gelling Technique The effect of the cooling of the agarose solution during mixing of cell samples was investigated in two experiments. A volume of cell suspension (0.5 mL) was injected into the agarose solution (1.5 mL) as usual, then the tube was allowed to cool in the air for a period of time before further mixing and pouring of the agarose mixture (Figure 5.2). Using an agarose solution at 40°C, the cooling of the gel over 20 seconds after initial mixing with cell samples caused apparent damage during subsequent vigorous mixing. There is a significant increase in tail moment (the significance of F is 0.03) and tail length with duration of cooling. There is not a significant increase in tail moment using a 55°C agarose stock solution (significance of F is 0.75). This suggests that some variation between samples is due to differences in sample manipulation when using cooler agarose solutions. From linear regression, the slope of the tail moment versus time for agarose cooling is 0.053 ± 0.014 iim per second for the 40°C agarose stock solution, and not significantly different from zero for 55°C agarose stock solution. 88 0.0 0.5 1.0 1.5 2.0 Time in Suspension (hours) Figure 5.1 Apparent damage due to cell suspens ion, a) Cells suspended- in PBS (filled symbols) show apparent damage compared to cel ls suspended in MEM or MEM + FBS (open symbols) . Symbols indicate the mean ± SEM of 50 comets per sl ide, b) The dif ference in apparent damage between cells suspended in PBS and cells suspended in other media. Symbols indicate the dif ference in means ± the sum of the SEMs of 50 comet per sl ide. The different symbols symbols denote different exper iments. 89 Figure 5.2 Apparent damage with cool ing of agarose during mixing. Separate exper iments used two t e m p -eratures of agarose solution (40°C o and 55°C • ) for unirradiated cell samples. Cells were added to agarose then allowed to cool for the indicate t ime before rapid mixing. Symbols indicate the mean ± SEM for 50 comets per sl ide 90 5.3.3 Duration of Alkaline Lysis and Rinse The data from three independent experiments measuring comets for different lysis times are shown in Figure 5.3. Large differences in apparent damage are present, suggesting that different levels of background damage may exist between experiments. Assuming the differences between results from different experiments are due to different levels of background damage, the difference in measured damage between 0.25 Gy and unirradiated samples for the same experiment may be used to eliminate the effect of initial damage, at least for the tail moment, since tail moment is linear with DNA damage over a wide range. The differences between 0 and 0.25 Gy comet values (a direct indication of assay sensitivity) do not show significant dependence on lysis duration (shown in Figure 5.3 c) and d)), where the slope of tail moment versus dose is not significantly different from zero (significance of F is 0.44). While not statistically significant, a decrease in apparent damage with increasing lysis time is seen. The effect of increasing durations of alkaline rinse for four independent experiments is shown in Figure 5.4. As for the lysis time data, large differences are seen between experiments which may be due to different levels of background damage. The difference between 0.25 Gy comets and unirradiated comets with different alkaline rinse duration is shown in Figure 5.4b), where the slope of the difference in tail moment versus rinse time is significant (0.79 ± 0.20 um per hour rinse time, significance level of F is 0.0036). The slope of the tail moment versus dose is also dependent on the apparent damage level of the unirradiated comets (Figure 5.5). The dependence of tail moment on the unirradiated comet tail moment shows a positive slope (0.61 ± 0.11 um per Gy) which is statistically significant (significance level of F is 0.0004). The effect of DNA damage level and rinse duration was investigated by measuring dose-response curves for two assay protocols using different rinse times. Two independent experiments were done for doses of 0 to 1 Gy with rinse times of 1 and 3 hours, and 1, 3 and 20 hours (Figure 5.6 and 5.8). The slope and offset of the dose response curves were 91 E CD O c CD l _ 0) a> E o 0 1 2 3 lysis duration (hours) lysis duration (hours) Figure 5.3 Apparent damage for varying lysis durat ion. Data f rom three experiments are indicated by different symbols , a) and b): Measurements are for cel ls i r r a d -iated with 0.25 Gy (fil led symbols) or 0 Gy (open symbols ) , c) and d): Tail length and moment d i f f e r -ences between 0.25 Gy and 0 Gy values for each e x p -er iment. Symbols are the mean ± SEM for tail length moment for a) and b). Symbols in c) and d) are the di f ferences in means ± the sum of the SEMs for these means. 92 Figure 5.4 Apparent damage due to rinse durat ion, a) , b): The effect of alkaline rinse duration on comet measurements for four separate experiments for 0 and 0.25 Gy irradiated cel ls . Symbols denote the mean ± SEM of 50 comets per sample, c ) , d): The di f ference between 0.25 Gy and 0 Gy data In a) and b). The symbols (open symbols are 0 Gy, f i l led symbols are 0.25 Gy) denote the dif ference in means ± the sum of the SEMs for these means. 93 2 3 4 5 6 0 Gy Tail Moment (/i-m) 7 Figure 5.5 The dependence of the di f ference between 0.25 Gy and 0 Gy comet measurements on background damage. The results of four separate experiments are shown using different symbols. Each symbol is the dif ference between the means of the 0.25 Gy and 0 Gy values. Error bars are the sum of the SEMs of the 0.25 Gy and 0 Gy means. 94 found from linear regression (Table 5.1, below). The differences between the comet values using 3 and 1 hour rinse protocols for the dose-response curves show an increase for increasing radiation dose, indicating that the effect of rinse duration on tail moment is dependent on the level of DNA damage (Figure 5.7). Further increases in rinse duration (up to 20 hours) cause further increasing apparent damage (Figure 5.8). (These are the same data at 1 and 3 hours rinse as Figure 5.6.) Two possibilities were considered to account for this dependence: the alkaline rinse is causing increased strand breaks, and the alkaline rinse is causing increased expression of DNA damage not due to the number of DNA strand breaks. Apparent damage in 0 and 1 Gy samples was examined in one experiment using rinse durations of 1 and 3 hours and rinse solutions containing the standard 0.03 M NaOH and 2 mM EDTA or 0.03 M NaOH alone. The apparent damage is higher in rinse solutions containing EDTA than in rinse solutions consisting of NaOH alone, regardless of radiation dose or rinse duration (Figure 5.9). The average variance of the tail moment values was also found to increase with alkaline rinse duration. Table 5.1 shows the standard deviation of tail moment values for the slides of each experiment, expressed as the mean of these values for slides of each experiment. The standard error of this mean is taken, as usual, as an estimate of the error of this mean. The assay resolution was evaluated as the standard deviation of tail moment values divided by the slope of the dose-response for tail moment versus dose. This value can be used to give an estimate of the separation in radiation dose necessary to give a statistically significant difference in means of measured tail moment. Assuming normal distributions of values, a difference of 2.33 standard errors gives a significance of 0.01 for the difference of means (two-tailed). Taking the standard error in the mean as the standard deviation divided by the square root of number of measurements, for a data set of 50 measurements the resolution of the assay (neglecting background damage due to other effects) is 0.08, 0.05 and 0.09 Gy for 1, 3 and 20 hour alkaline rinse durations. The size of the observed deviations from a linear fit (the residual) of the tail moment to the radiation 95 dose (when care was taken to avoid the identified sources of apparent damage) is also shown in Table 5.1. These represent the size of the scatter of the data due to background apparent damage and will ultimately limit the sensitivity of the assay. 96 Dose (cGy) D o s e ( c G y ) Figure 5.6 Dose response curves for 1 and 3 hour rinse durat ions. Open and closed symbols are for 1 and 3 hour alkaline rinse durat ions respect ively. Circ les and tr iangles are for separate exper iments. Symbols denote the mean ± SEM for 50 comets per sample . Lines for tail moment are f rom a l inear fit. 97 D > — T I - Q 1 20 ' — 1 — 1 — 1 — 1 1 1 1 1 j 1 ' 1 • 1 i—~—i i i i i i . i i i i L_ - 2 0 0 20 40 60 80 100 120 Dose (cGy) Figure 5.7 The dif ference in apparent damage d o s e -response due to alkaline rinse durat ion. Symbols denote the di f ferences between the tail moment data using 3 hour and 1 hour alkaline rinse durat ions (shown in Figure 5.6) for two exper iments. Symbols are the di f ferences in means ± the sum of the SEMs of these means. 98 120 Dose (cGy) D o s e ( c G y ) Figure 5.8 Dose response curves for rinse durat ions of 1, 3 and 20 hours. Data are from one experiment. Rinse durat ions are o 1 hour, • 3 hours, and v 20 hours. Symbols are the mean ± SEM for 50 comets per sample. 99 Figure 5.9 The effect of EDTA in alkaline rinse on apparent damage. Cells were irradiated with 0 or 1 Gy. The alkaline rinse const i tuents and duration are shown (0.03 M NaOH with or without 2 mM EDTA for 1 or 3 hours durat ion). Bars are the means ± SEM of 50 comets per sample. 100 Table 5.1 The Effect of Alkaline Rinse Duration on Tail Moment Measurements Rinse Duration (hours): Experiment 1 Experiment 2 Slope a (um/Gy) 1 1.8±0.3 3.6±0.3 3 7.7±0.6 9.1±0.6 20 N.P.c 11.4±2.1 Offseta (um) 1 2.27±0.15 3.16±0.17 3 4.56±0.32 6.13±0.31 20 N.P. 52.4±1.1 Standard Deviation b 1 0.91±0.14 1.96±0.21 (um) 3 1.92±0.74 3.24±0.40 20 N.P. 7.51±1.51 Standard Deviation 1 0.51±0.16 0.54±0.12 Slope 3 0.25±0.12 0.36±0.07 (Gy) 20 N.P. 0.66±0.25 Mean Residual d 1 0.21±0.15 0.27±0.10 (um) 3 0.42±0.10 0.42±0.43 Mean Residual 1 0.12±0.08 0.075±0.030 (eq.Gy) 3 0.055±0.030 0.046±0.046 b mean ± SEM of the standard deviations of the sets of comet measurements in the experiment, c N.P. denotes not performed. d the Mean Residual is the average of the absolute values of the differences between the data and the linear fit to the data. This value indicates (in um of tail moment or equivalent Gy) the average size of the scatter from the linear fit due to background apparent damage. 101 5.3.4 Summary of Apparent Damage Rates from Different Sources To compare the relative effects of changes in the assay protocol on the sensitivity for measuring DNA damage, the amount of apparent damage due to these protocol changes was converted to the equivalent level of radiation needed to produce these changes based on the slope of the appropriate dose-response curve. The results (Table 5.2) are expressed in equivalent Gy (ie. the amount of radiation in Gy that would yield the observed change in tail moment). The slope of the dose-response is assumed to be constant for these comparisons; and the uncertainty expressed for each value is calculated from the uncertainty in the change in tail moment with change in each protocol duration. Table 5.2 Apparent Damage Rates Due to Assay Protocol Timing Protocol Parameter Apparent Damage Rate a Cell suspension in PBS 0.055 ±0.019 eq. Gy per minute Agarose mixing (40°C agarose) 0.020 ± 0.005 eq. Gy per second Agarose mixing (55°C agarose) NS" Lysis time NS Rinse time c 1 hr rinse protocol, 0.05 Gy 0.0091 eq. Gy per minute 1 Gy 0.026 eq. Gy per minute 3 hour rinse protocol, 0.05 Gy 0.0031 eq. Gy per minute 1 Gy 0.0089 eq. Gy per minute a expressed in units ot equivalent Gy (the amount ot damage in Gy that would give this same value of tail moment). b NS denotes apparent damage rates that are not significantly different from zero. c The change in tail moment depends on both rinse time and radiation dose. These values are estimated from the change in tail moment for 10 extra minutes of alkaline rinse divided by 10. The error in these values is approximately 25% at these rinse durations and radiation doses. (Data not shown.) 102 5.4 Discussion The apparent damage due to storage of cell suspension in ice-cold PBS is the largest source of apparent damage that was investigated (Table 5.2). Since variations in this protocol duration are not expected to alter the assay response, this accumulating damage appears to be real. Since this damage accumulates at a rate of 0.055 Gy per minute, this source of damage would completely obscure damage due to low doses of radiation. Preparation of sets of slides typically takes 10 to 20 minutes depending on the number of slides; hence, all damage induced by radiation doses less than 1 Gy would be obscured. Apparent damage induction in cells suspended in warm PBS has been reported (McKelvey-Martin et al., 1993), but ice-cold PBS suspensions are commonly used and are not reported to result in apparent damage (Vijayalaxmi et al. 1992, for example). This raises the possibility that the PBS-induced damage observed here is due to other factors such as the presence of trypsin which is not inactivated by the growth serum during re-suspension of cells from the culture plate. This source of background damage was avoided by suspending cells in medium containing growth serum for all handling of cells. The level of apparent damage was found to be influenced by the temperature of the agarose stock solution and the duration of mixing of the cell suspension. This effect was avoided by using higher temperature agarose stock solution. As observed by workers using a similar assay (Vijayalaxmi et al., 1992) the expression of DNA damage was increased by prolonged exposure to alkali, an effect dependent on the radiation dose. Vijayalaxmi et al. (1992) suggest that this increase in measured damage is the result of conversion of alkali-labile sites to strand breaks. However, while this increase in apparent damage was observed here for increasing periods of alkaline rinse, it was not observed for increasing periods of alkaline lysis. The data, described here, from different experiments for increasing lysis duration show significant variability. This variability seems largely due to background damage which differed between experiments, since the differences in measured damage between 0.25 Gy and 103 unirradiated samples for the same experiment show much less variability. A slight decrease in measured damage occurs with increasing lysis duration but is not significant. The level of measured damage increases dramatically with increasing alkaline rinse duration. The large variation in measured damage between these experiments seems due to variations in background damage. The differences in measured damage between 0.25 Gy and unirradiated samples depends significantly on both alkaline lysis time and on the level of background damage in the unirradiated sample. The sensitivity of the assay depends on both the variability of the measured damage for a sample, and the separation between the averages. While the 3 hour alkaline rinse protocol increases the variability in measured damage for a sample, the increased slope of the dose-response curve compensates for this, yielding increased resolution as indicated by the minimum value of the standard deviation divided by the slope. If the increase in apparent damage following extended alkaline treatment is due to increased yield of DNA strand breaks by conversion of alkali labile sites, then this level of damage should correspond to the relative levels of immediate single-strand breaks (ssbs) and alkali labile sites. However, the level of alkali-labile sites is considered to be only 200-300 per Gy compared to 500-1000 immediate ssbs (Powell et al., 1990). Only if all base damage and sugar damage (as separate from alkali labile sites) are converted into ssbs can this increasing damage be explained by conversion of other damage to strand breaks by alkali treatment. Most significant is the dramatic increase in tail moment for unirradiated cells, which is 20 times higher for samples treated with a 20 hour alkaline rinse and 2 times higher for a 3 hour rinse, both compared to a 1 hour rinse. Increased apparent damage was not seen for increasing alkaline lysis duration. If alkaline rinse converts other damage types into ssbs, dependent only on the solution pH, then a constituent of the lysis solution (NaCl or N-lauroylsarcosine) would need to protect the DNA from this strand breakage. Since this seems unlikely, the increase in apparent damage for increasing rinse duration is apparently due to effects other than conversion of DNA damage into ssbs. 104 An alternative explanation for these data is that some relaxation of the DNA molecules occurs during rinse which affects the ability of the DNA to escape the comet head and form a tail during electrophoresis. This would explain the dependence of the tail moment on rinse duration and the lack of a plateau when all radiation induced damage should have been converted to ssbs. However, this view would not explain why this effect does not occur in the lysis solution since there is no increase in apparent damage for increasing lysis duration. This relaxation effect is apparently not unwinding of the DNA duplex since higher ionic strength enhances unwinding for DNA in solution. As well, since hydrophobic interactions are reported to dominate the DNA duplex stability in polar solvents (pp.116-159 in Saenger, 1984), the presence of detergent should decrease helix stability and promote strand separation. While the cause of these changes in damage expression with changes in alkaline rinse duration remains unclear, these data indicate that damage detection is enhanced by extended exposure to alkali. With the identified sources of apparent damage eliminated, the data still deviated significantly from a linear fit as indicated by the residuals shown in Table 5.1. These sources of apparent damage represent the greatest limitation of assay sensitivity. Since their cause is not known, the resolution of the assay can be brought below the size of these deviations (0.05 and 0.10 Gy, for the 3 and 1 hour rinse protocols) only by averaging results from several slides. The limits to resolution due to the spread in comet measurements may be sufficiently reduced by increasing the measurement sample size. 105 6. Conclusions The alkaline comet assay provides a sensitive means of quantitating single-strand breaks in the DNA of single mammalian cells. Factors limiting the sensitivity of the assay for measuring low levels of radiation-induced were examined. Image analysis software was written to improve detection of the comet image against the background fluorescence. Improvements were made by compensating for uneven response of the CCD camera by calibrating the camera response to zero illumination (dark-field subtraction). Flat-field correction (calibration of the camera response to a uniform, blank image) was not successful due to the production of unacceptable comet elements lying outside the proper area of the comet. Appropriate definitions of the comet edges were defined for software calculations and confirmed by direct observation. Since the image of the comet is identified based on a global threshold (elements with values three standard deviations above the mean of the background), the comet image depends strongly on the staining conditions. The fluorescent background could be suppressed by addition of 0.02 M NaCl to the staining solution and reduction of the concentration of the dye, propidium iodide, to as low as 0.025 iig/mL without compromising visibility of the comet when stained for several hours. Minimizing the stain concentration enhanced the comet visibility. But occasionally, the comets of some slides were found to be unusually dim if observed after storage, possibly due to exposure to distilled water from condensation in the storage container. A higher concentration of dye is therefore recommended to avoid this inconsistency between slides; 0.25 ug/mL was found to give good results. Fluorometric measurements of dye binding to DNA were in agreement with a model of two binding modes, both intercalated into the DNA duplex. The low affinity equilibrium constant corresponds well the binding of PI to DNA in solution. The high affinity equilibrium constant was 3-4 times higher than the high affinity equilibrium constant reported for dye binding to chromatin. 106 Several aspects of the assay protocol were identified which affect the sensitivity of the assay. Cells accumulated damage with time when suspended in PBS on ice but not when suspended in culture medium. However, since the use of cold PBS for cell suspensions in common, and is not reported to result in significant DNA damage, the reason for this increase in apparent damage is unclear. Low temperature (40°C) agarose stock solutions caused variability in damage measurements while higher temperature (55°C) solutions did not. Small differences in lysis duration did not cause variation in measured damage, while increasing alkaline rinse duration caused significant increases in measured damage. No satisfactory explanation for this increase in measured damage was determined. While the amount of measured damage increased to the highest rinse duration used, the width of the distribution of values also increased. Based on the width of the distribution of tail moment measurements for a sample and the slope of the dose-response of the tail moment and dose, the resolution of the assay was found to be approximately 0.05 Gy in experiments using a three hour alkaline rinse, for plateau-phase C3H IOIV2 cells and 50 measurements per sample. However, with the identified sources of apparent damage controlled, deviations from a linear dose-response occur at an average size equivalent to damage from 0.05 Gy. Unless additional sources of background damage can be identified, this limitation in the sensitivity of the alkaline comet assay cannot be overcome except by averaging measurements taken from replicate slides. 107 Bibliography Adams, R.P., Knowler, J.T., and Leader, D.P. The Biochemistry of the Nucleic Acids, 11th Edition. Chapman amd Hall, London, 1992. Angerer, L.M., and Moudrianakis, E.N. Interaction of Ethidium Bromide with Whole and Selectively Deproteinized Deoxynucleoproteins from Calf Thymus. J. Mole. Biol. 63, 505-521,1972. Bacq, A.M., Alexander, P. Fundamentals of Radiobiology. Pergamon Press, Bristol, 1961. Bauer, W. and Vinograd, J. The Interaction of Closed Circular DNA with Intercalative Dyes. J. Mole. Biol. 33, 141-171, 1968. Bernstein, C , and Bernstein, H. Aging, Sex and DNA Repair. Academic Press, San Diego, 1991. Bertuzzi, A., D'Angnano, I., Gandolfi, A., Graziano, A., Starace, G., and Ubezio, P. Study of Propidium Iodide Binding to DNA In Intact Cells by Flow Cytometry. Cell Biophysics 17, 257-267, 1990. Bloomfield, V.A., Crothers, D.M., Tinoco, I. Physical Chemistry of Nucleic Acids. Harper and Row, New York, 1974. Fairbairn, D.W., Olive, P.L., and O'Neill, K.L. The Comet Assay: A Comprehensive Review. Mutation Research, 339: 37-59, 1995. Frankenburg-Schwager, M. Induction, repair and biological relevance of radiation-induced DNA lesions in eukaryotic cells. Radiat. Environ. Biophys. 29, 273-292, 1990. Gonzalez, R.C, and Woods, R.C. Digital Image Processing. Addison-Wesley, Reading, Mass, 1992. Iliakis, G. The Role of DNA Double Strand Breaks in Ionizing Radiation-Induced Killing of Eukaryotic Cells. Bioessays 13, 641-648,1991. Johns, H.E., and Cunningham, J.R. The Physics of Radiology. Thomas Books, Springfield, Illinois, 1983. Kirkpatrick, F.H. Overview of Agarose Gel Properties. In Electrophoresis of Large DNA Molecules. Cold Spring Harbor Laboratory Press, New York, 1990. LePecq, J.B., and Paoletti, C. A Fluorescent Complex Between Ethidium Bromide and Nucleic Acids. J. Mole. Biol. 27, 87-106,1967. 108 Mackay,C.D. Charge-Coupled Devices in Astronomy. Ann. Rev. Astron. Astrophys. 24, 255-83,1986. Massey, P., Jacoby, G.H., CCD Data: The Good, The Bad, and The Ugly. In: Astronomical CCD Observing and Reduction Techniques. ASP Conference Series, Vol. 23, San Francisco, 1992. Muller, W.-U., Bauch, T., Streffer, C , Niedereichholz, F., and Bocker, W. Comet Assay Studies of Radiation-Induced DNA Damage and Repair in Various Tumour Cell Lines. Int. J. Radial Biol. 65, 315-319, 1994. Olive, P.L., Banath, J.P., Durand, R. Heterogeneity in Radiation-Induced DNA Damage and Repair in Tumor and Normal Cells Measured Using the "Comet" Assay. Rad. Res. 122, 86-94, 1990. Olive, P.L., Wlodek, D., Durand, R.E., and Banath, J.P. Factors Influencing DNA Migration From Individual Cells Subjected to Gel Electrophoresis. Exp. Cell Res. 198, 259-267, 1992. Olive, P.L., Banath, J.P., and Fjell, CD. DNA Strand Breakage and DNA Structure Influence Staining With Propidium Iodide Using the Alkaline Comet Assay. Cytometry 16, 305-312, 1994. Olive, P.L. and Banath, J.P. Radiation-Induced DNA Double-Strand Breaks Produced in Histone-Depleted Nuclei Measured Using the Neutral Comet Assay. Radiat. Res. 142: 144-152, 1995 Ostling, O., and Johanson, K.J. Microelectrophoretic Study of Radiation-Induced DNA Damage in Individual Mammalian Cells. Biochem. Biophys. Res. Com. 123, 291-298, 1984. Powell, S., McMillan, T.J. DNA Damage and Repair Following Treatment with Ionising Radiation. Radiotherapy and Oncology 19, 95-108, 1990. Press, W.H., Teukolsky, S.A., Vetterling, W.T., and Flannery, B.P. Numerical Recipes in C, 2nd Edition. Cambridge University Press, Cambridge, 1992. Prosperi, E., Giangare M . C , and Bottiroli, G. Nuclease-Induced DNA Structural Changes Assessed by Flow Cytometry with the Intercalating Dye Propidium Iodide. Cytometry 12, 323-329,1991. Rao, K.S., and Loeb, L.A. DNA Damage and Repair in Brain: Relationship to Aging. Mutation Research 275, 317-329, 1992. 109 Rydberg, B. The Rate of Strand Separation in Alkali of DNA of Irradiated Mammalian Cells. Rad. Res. 61, 274-287, 1975. Saenger, W. Principles of Nucleic Acid Structure. Springer-Verlag, New York, 1984. Singh, N.P., Stephens, R.E., and Scheider, E.L. Modifications of alkaline microgel electrophoresis for sensitive detection of DNA damage. Int. J. Radiat. Biol. 66, 23-28, 1994. Tice, R. and Setlow, R.B. DNA Repair and Replication in Aging Organisms and Cells. In: Handbook of the Biology of Aging, 2nd edition (C.E. Finch and E.L.Schneider, Eds.) Van Nostrand Reinhold Company, New York, 1985. Vijayalaxmi, Tice, R.R., Strauss, G.H.S. Assessment of radiation-induced DNA damage human blood lymphocytes using the single-cell gel electrophoresis technique. Mutation Research 271, 243-252,1992. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items