- Email: [email protected]
An automated new technique for scoring the rodent micronucleus assay: computerized image analysis of acridine orange supravitally stained peripheral blood cells
Fundamental and Molecular Mechanisms of Mutagenesis
ELSEVIER
Mutation Research 404 (1998) 149-154
An automated new technique for scoring the rodent micronucleus assay: computerized image analysis of acridine orange supravitally stained peripheral blood cells Norihide Asano a, *, Yoshiyuki Katsuma b, Hironobu Tamura c, Naohiko Higashikuni d, Makoto Hayashi e a Toxicological Research Division, Nitro Denko Corporation, 1-1-2, Shimohozumi, lbaraki, Osaka 567-8680, Japan b Biochemical Operation Department, Toyobo Co., Ltd., 2-2-8, Dojimahama, Kita-ku, Osaka 530-8230, Japan c Toxicology Department, Research Laboratories, Nippon Shinyaku Co., Ltd., Nishiohji Hachijo, Minami-ku, Kyoto 601-8550, Japan d General Research Institute, Itoham Foods Inc., 1-2 Kubogaoka, Moriya-cho, Kitasoma-gun, Ibaraki 302-0104, Japan e Division of Genetics and Mutagenesis, National Institute of Health Sciences, 1-18-1, Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan
Received 18 February 1998; accepted 22 February 1998
Abstract We developed an automated image analysis system to obtain objective data for the rodent peripheral blood micronucleus assay with acridine orange (AO) supravital staining. The system was able to identify micronucleated reticulocytes (MNRETs) and to evaluate inhibition of bone marrow cell proliferation by measuring the reticular area of reticulocytes (RETs). We also developed automated equipment to produce homogeneous acridine orange-coated glass slides. This study was designed to compare automated scoring with manual scoring using 4 model clastogens and 2 mouse strains. The MNRET incidence induced by each clastogen was similar for automated and manual scoring, and there was good correlation (r = 0.92) between the methods. In addition, an index of bone marrow toxicity based on the reticular area of RETs was compared to the conventional index (% of polychromatic erythrocytes (PCE) to total erythrocytes; PCE ratio) and was similar. The results indicated that our technique for computer-assisted image analysis for the micronucleus assay with AO supravitally stained peripheral blood RETs was comparable to conventional microscopic scoring, and it was superior in objectivity and statistical power. 9 1998 Elsevier Science B.V. All rights reserved. Keywords: Automatedmicronucleusassay; Image analysis;Peripheral blood; Acridine orange supravital staining
1. Introduction The micronucleus assay has been widely used to evaluate the clastogenicity of chemicals. In 1990,
* Corresponding author. Tel.: +81-726-21-0272; fax: +81726-21-0309; E-mail: [email protected]
Hayashi et al. [1] introduced a simple and reliable method to identify micronuclei in mouse peripheral reticulocytes (RETs) using an acridine orange (AO) supravital staining technique to distinguish D N A and RNA. The Collaborative Study Group for the Micronucleus Test (CSGMT) [2] validated the peripheral blood micronucleus assay with that staining
0027-5107/98/$19.00 9 1998 ElsevierScienceB.V. All rights reserved. PII: S0027-5 107(98)00108-0
150
N. Asano et al./ Mutation Research 404 (1998) 149-154
technique as an alternative to the mouse bone marrow micronucleus assay. C S G M T also validated the micronucleus assay in rat peripheral blood [3]. Manual microscopic evaluation of micronuclei is tedious and time-consuming and requires well-trained laboratory personnel, although it is faster and easier than metaphase analysis. To overcome those disadvantages and to obtain more objective data, automation techniques were developed both with flow cytometry [4-11] and with computerized image analysis [12-16]. Some of them required complex preparation procedures, however, prior to automation. We developed an automated micronucleus scoring system using computerized image analysis with newly developed software that effectively reduced 'noise' (nucleated cells, platelets, cell debris, etc.) and increased the ability to distinguish micronuclei and RETs based on their color, area, shape, and stain intensity. We used mouse peripheral blood and AO
pre-coated glass slides. This study compares our technique with manual scoring.
2. Materials and method 2.1. A u t o m a t i o n system
The automation system consisted of six independent components: a high performance personal computer, a 3CCD high-resolution color vision camera (SONY; XC-003), an auto-focus unit (Flovel; AF2000), a fluorescence microscope equipped with auto-scan unit (Olympus; BX60), a newly developed auto-slide exchanger, and a new image analyzing software developed by ourselves. The peripheral blood cells were captured through a 3CCD color vision camera into a personal computer as an image after autofocussing. The captured analogue image
Table 1 Comparison of automated and manual microscopic scoring of peripheral blood samples taken from mutagen-treatedmice and supravitally stained with acridine orange Chemical
Mouse strain
Dose (mg/kg)
MNRETs (%) mean 5: SD (Automated method)
MNRETs (%) mean 5: SD (Manual method)
BDFI
0.0 37.5 75.0 150.0 0.5
0.25 5:0.19 0.43 5:0.26 0.60 5:0.11 1.29 5:0.28 1.67 5:0.90
0.26 5:0.18 0.40 5:0.20 0.68 5:0.25 1.58 5:0.38 1.22 5:0.35
ddY
0.0 6.3 12.5 25.0 0.5
0.14 5:0.05 0.39 5:0.27 0.84 5:0.61 1.03 5:0.37 0.63 5:0.57
0.12 5:0.11 0.44 + 0.30 0.80 5:0.37 0.85 5:0.29 0.68 5:0.15
0.0 125.0 250.0 500.0 0.5
0.06 5:0.05 2.73 5:0.28 3.53 5:1.29 3.25 5:0.89 1.49 5:0.15
0.10 + 0.07 2.72 5:0.55 3.90 5:1.55 4.68 5:1.33 1.80 + 0.41
0.20 -I-0.11 0.72 5:0.25 2.05 5:1.04 3.47 5:0.96 1.93 _+0.89
0.16 + 0.11 0.73 5:0.29 1.83 5:0.81 3.22 + 1.57 1.86 + 0.63
2-AAF MMC
Ara-C MMC BDF1 BEN MMC ddY VCR MMC
0.0 0.0625 0.125 0.25 0.5
Abbreviations:2-AAF: 2-acetylaminofluorene,Ara-C: cytosine-fl-D-arabinofuranoside,BEN: benzene, VCR: vincristinesulfate, MNRETs: Micronucleatedreticulocytes.
151
N. Asano et al. /Mutation Research 404 (1998) 149-154
was converted to a RGB (red, green, and blue) digital image and identified as RETs and micronuclei separately by RGB relative intensity, followed by further noise reduction steps, i.e., analysis of area of target color, shape, and also sharpness of outline for micronucleus (MN). Finally, the MN image was superimposed on the RET image, and if the MN was on the RET, we identified a micronucleated reticulocyte (MNRET) and scored. The number of pixels of 'reticular area' for the reticulocytes were processed and recorded. 2.2. M i c r o n u c l e u s a s s a y
The micronucleus assay was performed using male BDF l (Charles River Japan, Atsugi, Japan) and ddY (SLC, Shizuoka, Japan) mice. Approximately seven-week-old mice were housed five per group at 2 2 + 2~ temperature, 55 + 10% humidity and, 12/12 h light/dark cycle, through the acclimation and experimental period. They were given commercial pellets and water ad libitum. 2-Acetylaminofluorene (2-AAF; Wako, Osaka, Japan), cytosine-fl-D-arabinofuranoside (Ara-C; Nippon Shinyaku, Kyoto, Japan), benzene (BEN; Wako)
and vincristine sulfate (VCR; Wako) were used as model micronucleus inducer, and mitomycin C (MMC; Kyowa Hakko, Tokyo, Japan) was used as a positive control in each assay. Dose levels of model chemicals were based on other papers [17-20]. After 48 h of single treatment, 5 /xl of peripheral blood was taken from the ventral tail, put on the center of an AO pre-coated glass slide and covered with a coverslip. The supravitally stained cells were analyzed both manually and automatically. We scored the number of MNRETs per 1000 RETs per mouse manually by fluorescence microscopy and at least 2000 RETs using automated analysis. We correlated the individual MNRET frequencies obtained by manual observation with those obtained by automated analysis. To obtain an index of bone marrow toxicity of VCR, we counted the number of RETs in 1000 erythrocytes.
3. R e s u l t s a n d d i s c u s s i o n
Automated and manual scoring produced similar results (Table 1), and a good correlation between the
8" y = 0.8x +0.2 r= 0.92
7
9
eDo//'o8~ _
i
9
OQ 9
0~ 0
9
J 1
i
i
i
i
2 3 4 5 MNRETs (%) Manual scoring
J 6
i 7
8
Fig. 1. Correlationbetween automatedanalysis and manual microscopicanalysis of MNRETs(included data for 2-AAF, Ara-C, BEN, VCR, and MMC).
152
N. Asano et al. / Mutation Research 404 (1998) 149-154
two methods was obtained ( r = 0.92, Fig. 1). The manual method occasionally gave slightly higher MNRET frequencies. The manual analysis might be subject to bias in scoring MNRETs in rather mature reticulocytes. MN induced by spindle poisons are large [21], and that was the case with VCR study. MN induced by model clastogens varied in pixel number, and many MN induced by VCR tended to contain a large number of pixels (Fig. 2).
Romagna and Staniforth [13] emphasized the importance of careful pre-treatment of samples when using bone marrow and May-Griinwald Giemsa staining for the micronucleus assay. The use of peripheral blood and AO supravital staining eliminates the need for pre-treatment, but high quality preparation is important for automated analysis. Adequate color intensity and well-spread blood cells are necessary to distinguish the MNRETs accurately. Therefore, we developed an automated method to
50 AAF (~128~cronucl~)
40 30 20 10 0
Ara-C (in 103 micronuclei)
40 30 20 ,m.i
10 0
ot..
40
ol.=l
30
BEN [ i n !fIQ m ~ n n ~ l t M ~
20 10 0 40
VCR (in 99 mic~onucLei)
30 20 10 0 40 30 20 10 0 O0
~
~ID
0
Image Size (Number of Pixels) Fig. 2. Distribution of MN size based on number of pixels; data shown for five clastogens.
153
N. Asano et al. / Mutation Research 404 (1998) 149-154
100 t ~n
1 "-
80
9
70
o ~ N'~
60
~
50
r o
[] Automated analysis with V C R [ ] Manual microscopic analysis with VCR
90"1
40 ~
30
'~
20 I0 0 0
0.062
0.125
0.250
MMC 0.5
Dose(mg/kg) Fig. 3. Comparison of computer generated and manually scored data for the number of reticulocytes with more than 15 red fluorescing pixels induced by the spindle poison vincristine.
coat glass slides with AO, and we used the machine-coated slides for this study. Such slides are now commercially available. The slides assured uniform high quality data in the automated assay. Autofocus is an important part of the system. Powerful focusing software overcome the difficulty of focusing the dark field. The system was able to distinguish MNRETs in the peripheral blood sample, and, with a slide auto-exchanger, a fully automated scoring of 30 slides could be conducted overnight. Bone marrow toxicity (PCE ratio) is hard to detect in the peripheral blood micronucleus assay with AO supravital staining because mature erythrocytes do not fluoresce. Iwakura et al. [18] noted that the proportion of type I-III RETs to total RETs (type I-IV) was a good indicator of bone marrow toxicity. The automated system reported here classified type I through type IV RETs by scoring red fluorescing reticulum pixels, and reticulocytes that had more than fifteen pixels were regarded as the target RETs (type I-III). The data generated by that technique correlated well with manually generated data (Fig. 3). Thus, this technique would be useful for detecting bone marrow toxicity in computerized
image analysis of the peripheral blood micronucleus assay. Acknowledgements
The authors are grateful to Dr. M. Bloom for reviewing this manuscript, and to Toyobo Co., Ltd., for supplying acridine orange pre-coated glass slides for this study. References [1] M. Hayashi, T. Morita, Y. Kodama, T. Sofuni, M. Ishidate Jr., The micronucleus assay with mouse peripheral blood reticulocytes using acridine orange-coated slides, Mutat. Res. 245 (1990) 245-249. [2] CSGMT, Micronucleus test with mouse peripheral blood erythrocytes by acridine orange supravital staining: The summary report of the 5th collaborative study by CSGMT/JEMS .MMS, Mutat. Res. 278 (1992) 83-98. [3] A. Wakata, Y. Miyamae, S. Sato, T. Suzuki, T. Morita, N. Asano, T. Awogi, K. Kondo, M. Hayashi, Evaluation of the rat micronucleus test with bone marrow and peripheral blood: The summary of the 9th collaborative study by CSGMT/MMS.JEMS, Environ. Mol. Mutagen. (1998) in press.
154
N. Asano et al. / Mutation Research 404 (1998) 149-154
[4] K.-J. Hutter, M. St~Shr, Rapid detection of mutagen induced micronucleated erythrocytes by flow cytometry, Histochemistry 75 (1982) 353-362. [5] M. Hayashi, H. Norppa, T. Sofuni, M. Ishidate Jr., Mouse bone marrow micronucleus test using flow cytometry, Mutagenesis 7 (1992) 251-256. [6] M. Hayashi, H. Norppa, T. Sofuni, M. Ishidate Jr., Flow cytometric micronucleus test with mouse peripheral erythrocytes, Mutagenesis 7 (1992) 257-264. [7] A.M. Tometsko, D.K. Torous, S.D. Dertinger, Analysis of micronucleated cells by flow cytometry: 1. Achieving high resolution with a malaria model, Mutat. Res. 292 (1993) 129-135. [8] A.M. Tometsko, S.D. Dertinger, D.K. Torous, Analysis of micronucleated cells by flow cytometry: 2. Evaluating the accuracy of high-speed scoring, Mutat. Res. 292 (1993) 137-143. [9] A.M. Tometsko, D.K. Torous, S.D. Dertinger, Analysis of micronucleated cells by flow cytometry: 3. Advanced technology for detecting clastogenic activity, Mutat. Res. 292 (1993) 145-153. [10] S.D. Dertinger, D.K. Torous, K.R. Tometsko, Flow cytometric analysis of micronucleated reticulocytes in mouse bone marrow, Mutat. Res. 390 (1997) 257-262. [11] J. Grawr M. Niisse, I.-D. Adler, Quantitative and qualitative studies of micronucleus induction in mouse erythrocytes using flow cytometry: I. Measurement of micronucleus induction in peripheral blood polychromatic erythrocytes by chemicals with known and suspected genotoxicity, Mutagenesis 12 (1997) 1-8. [12] A.P. Andrese, R.C. Shean, W.J. Kraft, Computer assisted microscopic enumerator and recording apparatus application in the mouse micronucleus test, Environ. Mutagen. 8 (1986) 5.
[13] F. Romagna, C.D. Staniforth, The automated bone marrow micronucleus test, Mutat. Res. 213 (1989) 91-104. [14] F. Vehaegen, A. Vral, J. Seuntjens, N.W. Schipper, L. de Ridder, H. Thierens, Scoring of radiation-induced micronuclei in cytokinesis-blocked human lymphocytes by automated image analysis, Cytometry 17 (1994) 119-127. [15] B.M. Schneider, F.E. Wiirgler, F. Romagna, Distinct area distribution differences of micronuclei induced by clastogenic and aneuploidogenic chemicals in the bone marrow of the CD-1 mouse, Mutat. Res. 334 (1995) 81-89. [16] J.W. Parton, W.P. Hoffman, M.L. Garriott, Validation of an automated image analysis micronucleus scoring system, Mutat. Res. 370 (1996) 65-73. [17] N. Asano, T. Hagiwara, The mouse peripheral blood micronucleus test with 2-acetylaminofluorene using the acridine orange supravital staining method, Mutat. Res. 278 (1992) 153-157. [18] K. Iwakura, H. Tamura, A. Matsumoto, S. Ajimi, S. Ogura, K. Kakimoto, T. Matsumoto, M. Hayashi, The micronucleus assay with peripheral blood reticulocytes by acridine orange supravital staining with l-/3-D-arabinofuranosylcytosine, Mutat. Res. 278 (1992) 131-137. [19] Y. Hatakeyama, E. Nakajima, H. Atai, S. Suzuki, Effects of benzene in a micronucleus test on peripheral blood utilizing acridine orange-coated slides, Mutat. Res. 278 (1992) 193195. [20] Y. Kondo, S. Honda, M. Nakajima, K. Miyahara, M. Hayashi, Y. Shinagawa, S. Sato, K. Inoue, S. Nito, F. Ariyuki, Micronucleus test with vincristine sulfate and colchicine in peripheral blood reticulocytes of mice using acridine orange supravital staining, Mutat. Res. 278 (1992) 187-191. [21] K.I. Yamamoto, Y. Kikuchi, A comparison of dia/neters of micronuclei induced by clastogens and by spindle poisons, Mutat. Res. 71 (1980) 127-131.
ELSEVIER
Mutation Research 404 (1998) 149-154
An automated new technique for scoring the rodent micronucleus assay: computerized image analysis of acridine orange supravitally stained peripheral blood cells Norihide Asano a, *, Yoshiyuki Katsuma b, Hironobu Tamura c, Naohiko Higashikuni d, Makoto Hayashi e a Toxicological Research Division, Nitro Denko Corporation, 1-1-2, Shimohozumi, lbaraki, Osaka 567-8680, Japan b Biochemical Operation Department, Toyobo Co., Ltd., 2-2-8, Dojimahama, Kita-ku, Osaka 530-8230, Japan c Toxicology Department, Research Laboratories, Nippon Shinyaku Co., Ltd., Nishiohji Hachijo, Minami-ku, Kyoto 601-8550, Japan d General Research Institute, Itoham Foods Inc., 1-2 Kubogaoka, Moriya-cho, Kitasoma-gun, Ibaraki 302-0104, Japan e Division of Genetics and Mutagenesis, National Institute of Health Sciences, 1-18-1, Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan
Received 18 February 1998; accepted 22 February 1998
Abstract We developed an automated image analysis system to obtain objective data for the rodent peripheral blood micronucleus assay with acridine orange (AO) supravital staining. The system was able to identify micronucleated reticulocytes (MNRETs) and to evaluate inhibition of bone marrow cell proliferation by measuring the reticular area of reticulocytes (RETs). We also developed automated equipment to produce homogeneous acridine orange-coated glass slides. This study was designed to compare automated scoring with manual scoring using 4 model clastogens and 2 mouse strains. The MNRET incidence induced by each clastogen was similar for automated and manual scoring, and there was good correlation (r = 0.92) between the methods. In addition, an index of bone marrow toxicity based on the reticular area of RETs was compared to the conventional index (% of polychromatic erythrocytes (PCE) to total erythrocytes; PCE ratio) and was similar. The results indicated that our technique for computer-assisted image analysis for the micronucleus assay with AO supravitally stained peripheral blood RETs was comparable to conventional microscopic scoring, and it was superior in objectivity and statistical power. 9 1998 Elsevier Science B.V. All rights reserved. Keywords: Automatedmicronucleusassay; Image analysis;Peripheral blood; Acridine orange supravital staining
1. Introduction The micronucleus assay has been widely used to evaluate the clastogenicity of chemicals. In 1990,
* Corresponding author. Tel.: +81-726-21-0272; fax: +81726-21-0309; E-mail: [email protected]
Hayashi et al. [1] introduced a simple and reliable method to identify micronuclei in mouse peripheral reticulocytes (RETs) using an acridine orange (AO) supravital staining technique to distinguish D N A and RNA. The Collaborative Study Group for the Micronucleus Test (CSGMT) [2] validated the peripheral blood micronucleus assay with that staining
0027-5107/98/$19.00 9 1998 ElsevierScienceB.V. All rights reserved. PII: S0027-5 107(98)00108-0
150
N. Asano et al./ Mutation Research 404 (1998) 149-154
technique as an alternative to the mouse bone marrow micronucleus assay. C S G M T also validated the micronucleus assay in rat peripheral blood [3]. Manual microscopic evaluation of micronuclei is tedious and time-consuming and requires well-trained laboratory personnel, although it is faster and easier than metaphase analysis. To overcome those disadvantages and to obtain more objective data, automation techniques were developed both with flow cytometry [4-11] and with computerized image analysis [12-16]. Some of them required complex preparation procedures, however, prior to automation. We developed an automated micronucleus scoring system using computerized image analysis with newly developed software that effectively reduced 'noise' (nucleated cells, platelets, cell debris, etc.) and increased the ability to distinguish micronuclei and RETs based on their color, area, shape, and stain intensity. We used mouse peripheral blood and AO
pre-coated glass slides. This study compares our technique with manual scoring.
2. Materials and method 2.1. A u t o m a t i o n system
The automation system consisted of six independent components: a high performance personal computer, a 3CCD high-resolution color vision camera (SONY; XC-003), an auto-focus unit (Flovel; AF2000), a fluorescence microscope equipped with auto-scan unit (Olympus; BX60), a newly developed auto-slide exchanger, and a new image analyzing software developed by ourselves. The peripheral blood cells were captured through a 3CCD color vision camera into a personal computer as an image after autofocussing. The captured analogue image
Table 1 Comparison of automated and manual microscopic scoring of peripheral blood samples taken from mutagen-treatedmice and supravitally stained with acridine orange Chemical
Mouse strain
Dose (mg/kg)
MNRETs (%) mean 5: SD (Automated method)
MNRETs (%) mean 5: SD (Manual method)
BDFI
0.0 37.5 75.0 150.0 0.5
0.25 5:0.19 0.43 5:0.26 0.60 5:0.11 1.29 5:0.28 1.67 5:0.90
0.26 5:0.18 0.40 5:0.20 0.68 5:0.25 1.58 5:0.38 1.22 5:0.35
ddY
0.0 6.3 12.5 25.0 0.5
0.14 5:0.05 0.39 5:0.27 0.84 5:0.61 1.03 5:0.37 0.63 5:0.57
0.12 5:0.11 0.44 + 0.30 0.80 5:0.37 0.85 5:0.29 0.68 5:0.15
0.0 125.0 250.0 500.0 0.5
0.06 5:0.05 2.73 5:0.28 3.53 5:1.29 3.25 5:0.89 1.49 5:0.15
0.10 + 0.07 2.72 5:0.55 3.90 5:1.55 4.68 5:1.33 1.80 + 0.41
0.20 -I-0.11 0.72 5:0.25 2.05 5:1.04 3.47 5:0.96 1.93 _+0.89
0.16 + 0.11 0.73 5:0.29 1.83 5:0.81 3.22 + 1.57 1.86 + 0.63
2-AAF MMC
Ara-C MMC BDF1 BEN MMC ddY VCR MMC
0.0 0.0625 0.125 0.25 0.5
Abbreviations:2-AAF: 2-acetylaminofluorene,Ara-C: cytosine-fl-D-arabinofuranoside,BEN: benzene, VCR: vincristinesulfate, MNRETs: Micronucleatedreticulocytes.
151
N. Asano et al. /Mutation Research 404 (1998) 149-154
was converted to a RGB (red, green, and blue) digital image and identified as RETs and micronuclei separately by RGB relative intensity, followed by further noise reduction steps, i.e., analysis of area of target color, shape, and also sharpness of outline for micronucleus (MN). Finally, the MN image was superimposed on the RET image, and if the MN was on the RET, we identified a micronucleated reticulocyte (MNRET) and scored. The number of pixels of 'reticular area' for the reticulocytes were processed and recorded. 2.2. M i c r o n u c l e u s a s s a y
The micronucleus assay was performed using male BDF l (Charles River Japan, Atsugi, Japan) and ddY (SLC, Shizuoka, Japan) mice. Approximately seven-week-old mice were housed five per group at 2 2 + 2~ temperature, 55 + 10% humidity and, 12/12 h light/dark cycle, through the acclimation and experimental period. They were given commercial pellets and water ad libitum. 2-Acetylaminofluorene (2-AAF; Wako, Osaka, Japan), cytosine-fl-D-arabinofuranoside (Ara-C; Nippon Shinyaku, Kyoto, Japan), benzene (BEN; Wako)
and vincristine sulfate (VCR; Wako) were used as model micronucleus inducer, and mitomycin C (MMC; Kyowa Hakko, Tokyo, Japan) was used as a positive control in each assay. Dose levels of model chemicals were based on other papers [17-20]. After 48 h of single treatment, 5 /xl of peripheral blood was taken from the ventral tail, put on the center of an AO pre-coated glass slide and covered with a coverslip. The supravitally stained cells were analyzed both manually and automatically. We scored the number of MNRETs per 1000 RETs per mouse manually by fluorescence microscopy and at least 2000 RETs using automated analysis. We correlated the individual MNRET frequencies obtained by manual observation with those obtained by automated analysis. To obtain an index of bone marrow toxicity of VCR, we counted the number of RETs in 1000 erythrocytes.
3. R e s u l t s a n d d i s c u s s i o n
Automated and manual scoring produced similar results (Table 1), and a good correlation between the
8" y = 0.8x +0.2 r= 0.92
7
9
eDo//'o8~ _
i
9
OQ 9
0~ 0
9
J 1
i
i
i
i
2 3 4 5 MNRETs (%) Manual scoring
J 6
i 7
8
Fig. 1. Correlationbetween automatedanalysis and manual microscopicanalysis of MNRETs(included data for 2-AAF, Ara-C, BEN, VCR, and MMC).
152
N. Asano et al. / Mutation Research 404 (1998) 149-154
two methods was obtained ( r = 0.92, Fig. 1). The manual method occasionally gave slightly higher MNRET frequencies. The manual analysis might be subject to bias in scoring MNRETs in rather mature reticulocytes. MN induced by spindle poisons are large [21], and that was the case with VCR study. MN induced by model clastogens varied in pixel number, and many MN induced by VCR tended to contain a large number of pixels (Fig. 2).
Romagna and Staniforth [13] emphasized the importance of careful pre-treatment of samples when using bone marrow and May-Griinwald Giemsa staining for the micronucleus assay. The use of peripheral blood and AO supravital staining eliminates the need for pre-treatment, but high quality preparation is important for automated analysis. Adequate color intensity and well-spread blood cells are necessary to distinguish the MNRETs accurately. Therefore, we developed an automated method to
50 AAF (~128~cronucl~)
40 30 20 10 0
Ara-C (in 103 micronuclei)
40 30 20 ,m.i
10 0
ot..
40
ol.=l
30
BEN [ i n !fIQ m ~ n n ~ l t M ~
20 10 0 40
VCR (in 99 mic~onucLei)
30 20 10 0 40 30 20 10 0 O0
~
~ID
0
Image Size (Number of Pixels) Fig. 2. Distribution of MN size based on number of pixels; data shown for five clastogens.
153
N. Asano et al. / Mutation Research 404 (1998) 149-154
100 t ~n
1 "-
80
9
70
o ~ N'~
60
~
50
r o
[] Automated analysis with V C R [ ] Manual microscopic analysis with VCR
90"1
40 ~
30
'~
20 I0 0 0
0.062
0.125
0.250
MMC 0.5
Dose(mg/kg) Fig. 3. Comparison of computer generated and manually scored data for the number of reticulocytes with more than 15 red fluorescing pixels induced by the spindle poison vincristine.
coat glass slides with AO, and we used the machine-coated slides for this study. Such slides are now commercially available. The slides assured uniform high quality data in the automated assay. Autofocus is an important part of the system. Powerful focusing software overcome the difficulty of focusing the dark field. The system was able to distinguish MNRETs in the peripheral blood sample, and, with a slide auto-exchanger, a fully automated scoring of 30 slides could be conducted overnight. Bone marrow toxicity (PCE ratio) is hard to detect in the peripheral blood micronucleus assay with AO supravital staining because mature erythrocytes do not fluoresce. Iwakura et al. [18] noted that the proportion of type I-III RETs to total RETs (type I-IV) was a good indicator of bone marrow toxicity. The automated system reported here classified type I through type IV RETs by scoring red fluorescing reticulum pixels, and reticulocytes that had more than fifteen pixels were regarded as the target RETs (type I-III). The data generated by that technique correlated well with manually generated data (Fig. 3). Thus, this technique would be useful for detecting bone marrow toxicity in computerized
image analysis of the peripheral blood micronucleus assay. Acknowledgements
The authors are grateful to Dr. M. Bloom for reviewing this manuscript, and to Toyobo Co., Ltd., for supplying acridine orange pre-coated glass slides for this study. References [1] M. Hayashi, T. Morita, Y. Kodama, T. Sofuni, M. Ishidate Jr., The micronucleus assay with mouse peripheral blood reticulocytes using acridine orange-coated slides, Mutat. Res. 245 (1990) 245-249. [2] CSGMT, Micronucleus test with mouse peripheral blood erythrocytes by acridine orange supravital staining: The summary report of the 5th collaborative study by CSGMT/JEMS .MMS, Mutat. Res. 278 (1992) 83-98. [3] A. Wakata, Y. Miyamae, S. Sato, T. Suzuki, T. Morita, N. Asano, T. Awogi, K. Kondo, M. Hayashi, Evaluation of the rat micronucleus test with bone marrow and peripheral blood: The summary of the 9th collaborative study by CSGMT/MMS.JEMS, Environ. Mol. Mutagen. (1998) in press.
154
N. Asano et al. / Mutation Research 404 (1998) 149-154
[4] K.-J. Hutter, M. St~Shr, Rapid detection of mutagen induced micronucleated erythrocytes by flow cytometry, Histochemistry 75 (1982) 353-362. [5] M. Hayashi, H. Norppa, T. Sofuni, M. Ishidate Jr., Mouse bone marrow micronucleus test using flow cytometry, Mutagenesis 7 (1992) 251-256. [6] M. Hayashi, H. Norppa, T. Sofuni, M. Ishidate Jr., Flow cytometric micronucleus test with mouse peripheral erythrocytes, Mutagenesis 7 (1992) 257-264. [7] A.M. Tometsko, D.K. Torous, S.D. Dertinger, Analysis of micronucleated cells by flow cytometry: 1. Achieving high resolution with a malaria model, Mutat. Res. 292 (1993) 129-135. [8] A.M. Tometsko, S.D. Dertinger, D.K. Torous, Analysis of micronucleated cells by flow cytometry: 2. Evaluating the accuracy of high-speed scoring, Mutat. Res. 292 (1993) 137-143. [9] A.M. Tometsko, D.K. Torous, S.D. Dertinger, Analysis of micronucleated cells by flow cytometry: 3. Advanced technology for detecting clastogenic activity, Mutat. Res. 292 (1993) 145-153. [10] S.D. Dertinger, D.K. Torous, K.R. Tometsko, Flow cytometric analysis of micronucleated reticulocytes in mouse bone marrow, Mutat. Res. 390 (1997) 257-262. [11] J. Grawr M. Niisse, I.-D. Adler, Quantitative and qualitative studies of micronucleus induction in mouse erythrocytes using flow cytometry: I. Measurement of micronucleus induction in peripheral blood polychromatic erythrocytes by chemicals with known and suspected genotoxicity, Mutagenesis 12 (1997) 1-8. [12] A.P. Andrese, R.C. Shean, W.J. Kraft, Computer assisted microscopic enumerator and recording apparatus application in the mouse micronucleus test, Environ. Mutagen. 8 (1986) 5.
[13] F. Romagna, C.D. Staniforth, The automated bone marrow micronucleus test, Mutat. Res. 213 (1989) 91-104. [14] F. Vehaegen, A. Vral, J. Seuntjens, N.W. Schipper, L. de Ridder, H. Thierens, Scoring of radiation-induced micronuclei in cytokinesis-blocked human lymphocytes by automated image analysis, Cytometry 17 (1994) 119-127. [15] B.M. Schneider, F.E. Wiirgler, F. Romagna, Distinct area distribution differences of micronuclei induced by clastogenic and aneuploidogenic chemicals in the bone marrow of the CD-1 mouse, Mutat. Res. 334 (1995) 81-89. [16] J.W. Parton, W.P. Hoffman, M.L. Garriott, Validation of an automated image analysis micronucleus scoring system, Mutat. Res. 370 (1996) 65-73. [17] N. Asano, T. Hagiwara, The mouse peripheral blood micronucleus test with 2-acetylaminofluorene using the acridine orange supravital staining method, Mutat. Res. 278 (1992) 153-157. [18] K. Iwakura, H. Tamura, A. Matsumoto, S. Ajimi, S. Ogura, K. Kakimoto, T. Matsumoto, M. Hayashi, The micronucleus assay with peripheral blood reticulocytes by acridine orange supravital staining with l-/3-D-arabinofuranosylcytosine, Mutat. Res. 278 (1992) 131-137. [19] Y. Hatakeyama, E. Nakajima, H. Atai, S. Suzuki, Effects of benzene in a micronucleus test on peripheral blood utilizing acridine orange-coated slides, Mutat. Res. 278 (1992) 193195. [20] Y. Kondo, S. Honda, M. Nakajima, K. Miyahara, M. Hayashi, Y. Shinagawa, S. Sato, K. Inoue, S. Nito, F. Ariyuki, Micronucleus test with vincristine sulfate and colchicine in peripheral blood reticulocytes of mice using acridine orange supravital staining, Mutat. Res. 278 (1992) 187-191. [21] K.I. Yamamoto, Y. Kikuchi, A comparison of dia/neters of micronuclei induced by clastogens and by spindle poisons, Mutat. Res. 71 (1980) 127-131.