Department of Preventive Medicine, Medical Research Center for Environmental Toxico-Genomics and Proteomics, College of Medicine, Korea University, Seoul 136705, Korea
Received March 11, 2004; accepted May 27, 2004
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key Words: Comet assay; quantile dispersion graphs; tail moment; Olive tail moment; extent tail moment; tail DNA; tail length; lymphocytes; DNA damage.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Although the popularity of the Comet assay in biomonitoring studies has increased (Moller et al., 2000), one of its shortcomings is a lack of agreement on a single appropriate Comet parameter that adequately describes DNA damage (Kassie et al., 2000
). Although there have been a few studies related to cell toxicity research, studies that compare Comet parameters in human biomonitoring studies are rare. Bocker et al. (1997)
reported that the tail moment and tail DNA showed more sensitivity than the tail length on the basis of an X-irradiation doseresponse experiment. De Boeck et al. (2000)
showed that tail DNA was a more appropriate parameter than tail length to analyze induced DNA damage because of its smaller variation, based on their internal standard. Olive et al. (1992)
have suggested that the tail DNA was more accurate for detecting DNA damages than the tail moment.
Meanwhile, the use of Comet assay for human studies has not shown consistent results. This is so because of its large variability as observed in both exposed and control groups (Rojas et al., 1999) or because of the poor dose-response relationship noted in biomonitoring studies (Moller et al., 2000
). It has also been found that the doseresponse correlations between benzene exposure and DNA damage in the exposed group were very differ from those in the control group (Sul et al., 2002
, 2003
). In case of B lymphocytes, which showed more significant damage in the exposed group than in the controls, the correlation coefficient between t,t-muconic acid levels and the tail moment was 0.737, and that for the tail DNA was 0.742. In the case of T lymphocytes, which showed a slight significant difference between the exposed group and the control group, the correlation coefficient was 0.345 for the tail moment and 0.319 for the tail DNA. They also reported that statistically significant DNA damage for tail parameters depended on cell types; the damage in B cells was higher than in T cells, G cells, and lymphocytes in the exposed group. These discrepancies imply that comparisons of tail parameters should be made according to the degree of DNA damage.
In this study, a new graphical approach, the so-called quantile dispersion graphs (QDGs), was used to compare tail parameters in both high and low DNA damaged cell types with human biomonitoring data. The QDGs were first introduced by Khuri (1997), and he employed exact quantiles of the distribution of the parameter of interest to compare the statistical estimation methods of variance components for random analysis of variance (ANOVA) models. When an exact distribution of the parameter of interest was unknown, Lee and Khuri (1999
, 2000)
suggested the use of empirical quantiles, which produced the so-called empirical quantile dispersion graphs (EQDGs). Such plots have been shown to provide not only a comprehensive assessment of the estimation capability of experimental designs but also an overall evaluation of the quality of statistical estimation methods (Khuri and Lee, 2003
). We employed this method to compare characteristics among the Comet parameters. The EQDGs consisted of plots of the minimum and maximum values of empirical quantiles of the tail parameters over a parameter space. These plots provided an assessment of overall quality of the tail parameters for a particular cell type in the Comet assay, because the plots utilized all of the human data obtained from Comet studies. Furthermore, the plots could be conveniently used to compare an exposed group with a control group for each of the tail parameters.
![]() |
Materials and Methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
All subjects are male, and their average age and smoking status are similar to the study population of Sul et al. (2002, 2003)
. Each of sampling times for the exposed and control groups was within a week.
Comet data and tail parameters. The Comet experiment for blood lymphocytes and granulocytes was performed by two laboratory experts, from July 2000 to October 2002, in the Medical Research Center for Environmental Toxico-Genomics and Proteomics, Korea University. The alkaline version of the Comet assay was performed according to the method of Singh et al. (1988), with only one minor modification (pH > 13). Two slides were prepared, and each of about 50 randomly chosen cells (a total of
100 cells, up to 120 cells) was scored manually. For scoring, a strict slide reading manual, which included a starting point for reading, a reading area in the slide, a reading direction, and click numbers to move was used. For sampling and scoring, a single-blind method was used. In brief, cells were mixed with 0.5% low-melting agarose and then placed onto fully frosted slides. After that, the slides were coated with 1% normal agarose. When the agarose had solidified, an additional layer of 0.5% low-melting agarose was added. The slides were submersed in the lysing solution (1.5 M NaCl; 100 mM EDTA-1Na; 10 mM Tris-HCl, pH 10; 1% Triton X-100, and 10% DMS, pH 10) for one hour. The slides were then placed in unwinding buffer (1 mM EDTA and 300 mM NaOH, pH > 13) for 20 min. Electrophoresis was carried out using the same solution for 20 min at 25 V and 300 mA (0.8 V/cm). After electrophoresis, the slides were neutralized by washing three times with neutralization buffer (400 mM Tris-Cl, pH 7.4) for 5 min each, and the cells were stained with 50 µl of 10 µg/ml ethidium bromide.
Tail parameters were then calculated automatically using the Komet 4.0 image analysis system (Kinetic Imaging, Liverpool, UK) fitted with an Olympus BX50 fluorescence microscope that was equipped with an excitation filter of 515560 nm and a barrier filter of 590 nm. Tail parameters used in this study were the tail moment (TM), the extent tail moment (ETM), the tail DNA % (TD), and the tail length (TL). The TM was defined by the percentage of DNA in the tail multiplied by the length between the center of the head and tail, which was defined by Olive et al. (1990a). The ETM was a product of the tail length and the TD. Note that in some studies the extent tail moment has been expressed as the tail moment (De Boeck et al., 2000
; Lee and Steinert, 2003
; Tice et al., 2000
).
The Comet assay was carried out within 3 h after blood sampling. Mononuclear and polynuclear cells were separated by using two kinds of histopaque solutions (1.077 and 1.119) with centrifugation at 700 g for 30 min at room temperature for the sampling of lymphoctytes and granulocytes. T lymphocytes, B lymphocytes, and granulocytes were positively selected with magnetic beads (magnetic cell sorting (MACS) CD3 or CD19 or CD15 isolation kit; Miltenyi Biotec) according to the manufacturer's instructions. Cell viability was measured by the uptake of trypan blue. The viabilities of lymphocytes were 97.6 ± 0.6% after 2 h of sample preparation, 95.1 ± 0.3% after 3 h, and 94.5 ± 0.7% after 4 h. The methods for the Comet assay and cell preparation have been described in detail elsewhere (Sul et al., 2002, 2003
).
Statistical methods. Noting that the observed values of DNA damage for each subject show a skewed distribution, median values of DNA damage were selected to represent the amount of DNA damage. Comparisons of the averages of the median DNA damages between the exposed and control groups were made using the two-tailed Student's t-test. Quantiles of the observed values of DNA damage for each subject were used for a more comprehensive comparison of the tail parameters, namely, TM, ETM, TD, and TL. This was achieved by noting that the distribution of the amount of DNA damage could be determined in terms of its quantiles (Khuri and Lee, 1998). Considering that more than 100, and up to 120 Comet-assayed cells for each of the subjects were used, the use of quantiles of the observed DNA damage for a subject would more appropriately represent the true status of DNA damage, rather than using a single-valued criteria, such as a mean or a median. Furthermore, the pattern of such quantiles over the exposed or control groups would provide overall information about DNA damage for the entire group.
As mentioned earlier, for a given tail parameter, the dependency of the particular empirical quantiles on the value of the true amount of DNA damage can be assessed by computing the minimum and maximum of its empirical quantiles over the study subjects. Plots of the resulting minima and maxima produce the EQDGs (Lee and Khuri, 1999). More specifically, let
ijk be an observed extent of DNA damage for an ith parameter of a jth cell in a kth subject, i = TM, ETM, TD, and TL; j = 1, 2, ..., nk; and k = 1, 2, ..., mE (or mC), where nk is the number of cells observed in the kth subject and mE (or mC) is the number of subjects for the exposed (or the control) group. (Note that in our study, mE = 259, mC = 235, nk = 100
120.) Also, let
ik be a set of such values observed for the kth subject having ith parameter, namely,
ik = {
i1k,
i2k, ...,
i,nk,k}. For the given parameter, i, and the subject, k, let q(p,
ik) denote the pth quantile of
ik, that is, p(
ik
q) = p, 0
p
1. Define
![]() |
![]() |
Plotting qmin(p, i) and qmax(p,
i) against selected values of p (0
p
1) provides the EQDGs for the ith parameter, i = TM, ETM, TD, and TL. These EQDGs will provide information about the distribution of DNA damage for each of the tail parameters. In a comparison of the exposed group with the control group, the ith parameter is considered to be "efficient" to assess the amount of DNA damage if values of both qmax(p,
i) and qmin(p,
i) for the former group are consistently higher, over the range of p, than those for the latter group.
To compare the tail parameters, however, we have to standardize the values of ijk, because of its differing ranges for the tail parameters. For each of an ith parameter, i = TM, ETM, TD, and TL, we transformed
ijk's to
![]() |
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
The minimum and maximum empirical quantile values of the observed DNA damage for selected probability values for each of the cell types are shown in Tables 25. The EQDG plots for the lymphocytes, the T cells, the B cells, and the G cells are given in Figures 1, 2, 3, and 4, respectively. For lymphocyte cells, the exposed group showed consistently higher maximum quantile values [i.e., qmax(p,
i)], over the range of p (0
p
1) than did the control group in the TM. In the TD, the maximum quantile values for the exposed group are mostly higher than those for the control group (see Table 2 as well as Figure 1a and 1c). However, the maximum quantiles for the ETM and the TL between the exposed group and the control group are crossed over. For the minimum quantiles [i.e., qmin(p,
i)], the control group shows slightly higher values than the exposed group for most values of p for all parameters (see Fig. 1). For T-cells, most of the maximum quantiles of the TM are larger for the exposed group than for the control group (see Table 3 and Fig. 2a). The exposed group's preference to the control group in producing larger quantiles, however, is unclear for the parameters ETM and TD (see Fig. 2b and 2c). With the TL, consistently higher quantiles are observed, even in the control group (see Fig. 1d, Fig. 2d and Fig. 4d). For B cells, the maximum quantiles in the exposed group are uniformly higher than those for the control group at all tail parameters. This is also true for the minimum quantiles, except for the TL (see Table 4 and Fig. 3). For G cells, the TM, ETM, and TD parameters in the exposed group show larger maximum quantiles than those in the control group, but this pattern is reversed with the TL (see Table 5 and Fig. 4). In conclusion, in a comparison of the exposed and control groups, either the TM or the TD is a more efficient parameter than the others for the lymphocyte, T cells and G cells. For B cells, all parameters, except the TL, are efficient.
|
|
|
|
|
|
|
|
|
|
|
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Several parameters of Comet features have been developed. Bocker et al. (1997) summarized 10 measurement methods, and a new parameter, the tail profile, has been introduced by Bowden et al. (2003)
. However, the TM, TD, and TL remain the most frequently used tail parameters (De Boeck et al., 2000
). Among them, the TL was reported as the least sensitive parameter in the doseresponse curve when compared with the TM and a head-tail ratio (Bocker et al., 1997
). De Boeck et al. (2000)
have also reported that the TL showed more inter-electrophoresis and inter-experimenter variation than the TD.
Even for healthy individuals, the variation of tail parameters between studies was considerable (1.2-fold to 26-fold) (Moller et al., 2000). The tail moments of the exposed and control groups in this study, however, were relatively similar to other benzene-exposed groups, as measured with Komet 4.0 software (Zhu et al., 2000
). The average median of the tail moments of exposed workers ranges from 1.17 to 1.27 in our study, whereas the mean of the tail moments in the study of Zhu et al. (2000)
for lymphocytes ranged from 1.47 to 2.25. For the controls, the values were 1.121.22 in our study and 1.39 in Zhu's study. Andreoli et al. (1997)
have shown similar results; they report that the tail moment of the exposed group was 1.90 and that of the control group was 0.94. Lam et al. (2002)
, however, showed lower values than ours; 0.74 for exposed subjects and 0.53 for controls.
Previous studies have shown that the parameters of TM and TD were generally highly correlated. The correlation between the TM and the TL of human lymphocytes was relatively lower than between the TM and the head-tail ratio, both in normal subjects and in ataxia telangiectasia patients (Bocker et al., 1997). It has also been reported that the second derivatives of the TM and those of the TD in the dosetimeresponse surface, which quantified DNA damage/repair, were highly correlated (Kim et al., 2002
). For our data, the correlations among the four parameters are summarized in Table 6. Those between the TM, or the ETM, and the TD are generally higher than those for the TM and the TL, or the TD and the TL, in both the exposed and control groups. Over the cell types considered, the correlation coefficients between the TD and the TL are less than 0.2 in the control group, and between 0.2 and 0.7 in the exposed group. Although the correlation coefficients between the TM, or the ETM, and the TD are similar in both groups, those between the TM and the TL, or between the TD and the TL, are very different in both groups. The correlations between the ETM and the TL are also noted to be higher than those between the TM and the TL, because the ETM uses a tail length, whereas the TM uses a distance between the center of the head and the tail.
|
In the EQDGs for all cell types we considered, only the TM shows consistently higher maximum quantile values in the exposed group than in the control group. Moreover, although the minimum quantile values of the TM and the ETM are almost unchanged over the range of probabilities, the values for the TD increase noticeably for large probability values, regardless of cell type. In this respect, the TM is a more efficient parameter than the others in comparing the exposed and control groups. For the B cells, which represent high levels of damage, however, the parameters of ETM and TD are also useful (see Figs. 14). When considering the consistency of quantiles for DNA damage, the TM is preferred for human biomonitoring, for all cell types and for both the high and low damage status, because it gives the least variable estimates of DNA damage. As for the degree of uniformity of the observed quantiles, the parameter TM is more stable than the others for the lymphocytes and B cells, whereas the ETM and the TD provide stable estimates in the B cells and G cells (see Figs. 5
8).
In conclusion, this study compared the Comet parameters of TM, ETM, TD, and TL in several ways; a comparison of means between the exposed group and the control group was made using Student's t-test. A relationship among those parameters was established using correlation analysis, and a more comprehensive evaluation of the parameters was done by using the so-called EQDGs. Considering all of the results from these analyses, it was shown that the TM is generally the preferred parameter among the four Comet parameters. Some other recommendations, however, have to be considered at this point. Tice et al. (2000) recommended that data on the TL and the TD should be provided when using the TM. De Boeck et al. (2000)
have also pointed out that the TM should be addressed carefully and always in addition to the TD and/or the TL, because of its masking effect in some cases. In the present study, the TD also showed significant results, but at a different perspective from those of the TM or the ETM (see Table 1). It would therefore be advisable to produce the TM and the TD at the same time by using a computerized image-analysis system, and to present both sets of information as the tail parameters for human biomonitoring studies.
![]() |
ACKNOWLEDGMENTS |
---|
![]() |
NOTES |
---|
1 To whom correspondence should be addressed at Department of Preventive Medicine, College of Medicine, Korea University, 126-1, 5-ka, Anam-Dong, Sungbuk-Gu, Seoul 136-705, Korea. Fax: + 82-2-927-7220. E-mail: jyleeuf{at}korea.ac.kr.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Andreoli, C., Leopardi, P., and Crebelli, R. (1997). Detection of DNA damage in human lymphocytes by alkaline single cell gel eletrophoresis after exposure to benzene or benzene metabolites. Mutat. Res. 377, 95104.[ISI][Medline]
Bajpayee, M., Dhawan, A., Parmar, D., Pandey, A. K., Mathur, N., and Seth, P. K. (2002). Gender-related differences in basal DNA damage in lymphocytes of a healthy Indian population using the alkaline Comet assay. Mutat. Res. 520, 8391.[ISI][Medline]
Bocker, W., Bauch, T., Muller, W. U., and Streffer, C. (1997). Image analysis of Comet assay measurements. Int. J. Radiat. Biol. 72, 449460.[CrossRef][ISI][Medline]
Bowden, R. D., Buckwalter, M. R., McBride, J. F., Johnson, D. A., Murray, B. K., and O'Neill, K. L. (2003). Tail profile: A more accurate system for analyzing DNA damage using the Comet assay. Mutat. Res. 537, 18.[ISI][Medline]
De Boeck, M., Touil, N., De Visscher, G., Vande, P. A., and Kirsch-Volders, M. (2000). Validation and implementation of an internal standard in Comet assay. Mutat. Res. 469, 181197.[ISI][Medline]
Garaj-Vrhovac, V., and Zeljezic, D. (2002). Assessment of genome damage in a population of Croatian workers employed in pesticide production by chromosomal aberration analysis, micronucleus assay and Comet assay. J. Appl. Toxicol. 22, 249255.[CrossRef][ISI][Medline]
Kassie, F., Parzefall, W., and Knasmuellers, S. (2000). Single cell gel electrophoresis assay: A new technique for human biomonitoring studies. Mutat. Res. 463, 1331.[ISI][Medline]
Khuri, A. I. (1997). Quantile dispersion graphs for analysis of variance estimates of variance components. J. Appl. Stat. 24, 711722.[CrossRef][ISI]
Khuri, A. I., and Lee, J. (1998). A graphical approach for evaluating and comparing designs for nonlinear models. Comp. Stat. Data. Anal. 27, 433443.[CrossRef][ISI]
Khuri, A. I., and Lee, J. (2003). Quantile dispersion graphs for comparing designs for random ANOVA models. Proceedings of Graybill Conference, 2001, on Linear, Nonlinear, and Generalized Linear Models. Department of Statistics, Colorado State University, Fort Collins, Colorado 80523, pp. 227240.
Kim, B. S., Park, J. J., Edler, L., Von Fournier, D., Haase, W., Sautter-Bihl, M. L., and Gotzes, F. (2002). New measure of DNA repair in the single-cell gel electrophoresis (Comet) assay. Environ. Mol. Mutagen. 40, 5056.[CrossRef][ISI][Medline]
Kohn, K. W. (1991). Principles and practice of DNA filter elution. Pharmacol. Ther. 49, 5577.[CrossRef][ISI][Medline]
Lam, T. H., Zhu, C. Q., and Jiang, C. Q. (2002). Lymphocyte DNA damage in elevator manufacturing workers in Guangzhou, China. Mutat. Res. 515, 147157.[ISI][Medline]
Lee, J., and Khuri, A. I. (1999). Graphical technique for comparing designs for random model. J. Appl. Stat. 26, 933947.[CrossRef][ISI]
Lee, J., and Khuri, A. I. (2000). Quantile dispersion graphs for the comparison of designs for a random two-way model. J. Stat. Plan. Infer. 91, 123137.[CrossRef][ISI]
Lee, R. F., and Steinert, S. (2003). Use of the single cell gel electrophoresis/Comet assay for detecting DNA damage in aquatic (marine and freshwater) animals. Mutat. Res. 544, 4364.[ISI][Medline]
Moller, P., Knudsen, L.E., Loft, S., and Waalin, H. (2000). The Comet assay as a rapid test in biomonitoring ocupational exposure to DNA damaging agents and effect of counfounding factors. Cancer Epidemiol. Biomarkers Prev. 9, 10051015.
Muller, W. U., Bauch, T., Streffer, C., Niedereichholz, F., and Bocker, W. (1994). Comet assay studies of radiation-induced DNA damage and repair in various tumor cell lines. Int. J. Radiat. Biol. 65, 315319.[ISI][Medline]
Olive, P. L., Banath, J. P., and Durand, R. E. (1990a). Heterogeneity in radiation induced DNA damage and repair in tumor and normal cells using the "Comet" assay. Radiat. Res. 122, 8694.[ISI][Medline]
Olive, P. L., Banath, J. P., and Durand, R. E. (1990b). Detection of etoposide resistance by measuring DNA damage in individual Chinese hamster cells. J. Natl. Cancer Inst. 82, 779783.[Abstract]
Olive, P. L., Wlodek, D., Durand, R. E., and Banath, J. P. (1992). Factors influencing DNA migration from individual cells subjected to gel electrophoresis. Exp. Cell. Res. 198, 259267.[ISI][Medline]
Ostling, O., and Johanson, K. J. (1984). Micro gel electrophoretic study of radiation induced DNA damages in individual mammalian cells. Biochem. Biophys. Res. Commun. 123, 291298.[ISI][Medline]
Rojas, E., Lopez, M. C., and Valverde, M. (1999). Single cell gel electrophoresis assay: Methodology and applications. J. Chromatogr. B 722, 225254.[CrossRef]
Schabath, M. B., Spitz, M. R., Grossman, H. B., Zhang, K., Dinney, C. P., Zheng, P. J., and Wu, X. (2003). Genetic instability in bladder cancer assessed by the Comet assay. J. Natl. Cancer. Inst. 95, 540547.
Singh, N. P., McCoy, M. T., Tice, R. R., and Schneider, E. L. (1988). A simple technique for quantification of low levels of DNA damage in individual cells. Exp. Cell. Res. 175, 184191.[ISI][Medline]
Sul, D., Lee, D., Im, H., Oh, E., Kim, J., and Lee, E. (2002). Single strand DNA breaks in T- and B-lymphocytes and granulocytes in workers exposed to Benzene. Toxicol. Lett. 134, 8795.[CrossRef][ISI][Medline]
Sul, D., Oh, E., Im, H., Yang, M., Kim, C-W., and Lee, E. (2003). DNA damage in T- and B-lymphocytes and granulocytes in emission inspection and incineration workers exposed to polycyclic aromatic hydrocarbons. Mutat. Res. 538, 109119.[ISI][Medline]
Tice, R. R., Agurell, E., Anderson, D., Burlinson, B., Hartmann, A., Kobayashi, H., Muyamae, Y., Rojas, E., Ryu, J. C., and Sasaki, Y. F. (2000). Single cell gel/Comet assay: Guideline for in vitro and in vivo genetic toxicology testing. Environ. Mol. Mutagen. 35, 206221.[CrossRef][ISI][Medline]
Zhu, C. Q., Lam, T.H., Jiang, C. Q., Wei, B. X., Xu, Q. R., and Chen, Y. H. (2000). Increased lymphocyte DNA strand breaks in rubber workers. Mutat. Res. 470, 201209.[Medline]