Mutagenesis, Vol. 16, No. 3, 189-195,
May 2001
© 2001 UK Environmental Mutagen Society/Oxford University Press
Evaluation of inter-scorer and inter-laboratory reliability of the mouse epididymal sperm aneuploidy (m-ESA) assay
Institut für Säugetiergenetik, GSF-Forschungszentrum für Umwelt und Gesundheit GmbH, Neuherberg, Ingolstädter Landstraße 1, D-85764, Neuherberg, Germany, 1 Biology and Biotechnology Research Program, Lawrence Livermore National Laboratory, PO Box 808, 7000 East Avenue, Livermore, CA 94550, USA, 2 Laboratory of Toxicology, National Institute of Environmental Health Sciences, Research Triangle Park, NC 27709, USA and 3 BioStatistics Branch, National Institute of Environmental Health Sciences, Research Triangle Park, NC 27709, USA
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The mouse epididymal sperm aneuploidy (mESA) assay using 3-chromosome fluorescence in situ hybridization (FISH) was recently developed for assessing the aneugenic potential of chemicals on male germ cells. This study was designed to identify the major technical factors that affect inter-scorer and inter-laboratory variability of the mESA assay. Two laboratories participated in this study (GSF and Lawrence Livermore National Laboratory, LLNL). Mice (102/ElxC3H/El) F1 were exposed in one laboratory (GSF) to vinblastine (VBL; single intraperitoneal injection of 0, 0.5, 1.0 or 2.0 mg/kg), one of the 10 priority compounds of the Commission of the European Communities (CEC) Aneuploidy Program. Twenty-two days later the mESA assay was applied to analyze sperm aneuploidy. In the initial evaluation, small but statistically significant differences were found between the two laboratories in baseline frequencies and there was also disagreement in the determination of a VBL aneuploid effect. Therefore, experiments were conducted to identify the sources of the inter-laboratory differences and technical factors that affected assay reliability and the VBL study was repeated. A harmonization experiment was conducted by bringing the microscope scorers from both laboratories to the same site (LLNL) for a cross-training exercise. Following this exercise, a second group of VBL-treated and control mice were evaluated, and we concluded that VBL is not a sperm aneugen. Our research has identified scoring criteria as the major source of inter-laboratory variation and emphasizes the importance of strict technical controls for the mESA assay, including controlling slide preparations for treatment-induced reductions in sperm count, coding of slides and selection of statistical tests. These considerations are particularly important for the interpretation of small effects (
2-fold) on sperm aneuploidy. Our findings suggest that 2-fold differences in frequencies can result from differences among scorers, samples and treatment groups, and are readily within the normal variation for the mESA assay. Such small differences should be viewed with caution until independently confirmed. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Aneuploidy transmitted via germ cells leads to several common genetic disorders in humans. Although most aneuploid conceptions are lost prior to birth (Hassold, 1986
), ~1/300 newborns are affected (Jacobs, 1992). The relative parental contribution to germinal aneuploidy varies by chromosome type, with autosomes showing major maternal contributions and sex chromosomes showing substantial paternal contributions (Wyrobek, 1993). However, little is known about risk factors for germ cell aneuploidy (physiological and genetic factors, or chemical agents).
Fluorescence in situ hybridization (FISH) with chromosome-specific DNA probes has been successful in detecting human sperm carrying abnormal numbers of chromosomes, such as hyperhaploidy, hypohaploidy and diploidy (Baumgartner et al., 1999; Robbins et al., 1993, 1995, 1997a; Spriggs et al., 1995). Increased incidences of aneuploid sperm have been reported among cancer patients after they received chemotherapy (Robbins et al., 1997a), among active cigarette smokers (Rubes et al., 1998) and among men who consumed caffeine and alcohol (Robbins et al., 1997b). Given the difficulties in conducting human exposure studies, especially those with complex exposures, reliable experimental animal models are needed to identify potential germ cell aneugens and to characterize their mechanisms of action.
Wyrobek et al. (1995) developed a mouse model using multi-color FISH to detect aneuploidy in testicular sperm (mTSA). This mTSA assay was used to demonstrate increased sperm aneuploidy with advanced male age (Lowe et al., 1995), and elevated frequencies of aneuploid sperm in certain translocation carriers (Baulch et al., 1996). Lowe et al. (1996) extended this approach to the analysis of the homogeneous pool of epididymal sperm by developing the mouse epididymal sperm aneuploidy (mESA) assay. These authors then applied it to measure baseline frequencies of aneuploidy in sperm from healthy adult males of several strains of mice as well as from mice carrying Robertsonian translocations (Lowe et al., 1996). The inter-laboratory comparison study of Adler et al. (1996) noted small differences in baseline frequencies of sperm aneuploidy between the two collaborating laboratories (GSF and Lawrence Livermore National Laboratory, LLNL) that were within a factor of two (P > 0.05). In human studies using sperm FISH, even larger variations in baseline frequencies of abnormal sperm are commonly reported among laboratories (e.g., Downie et al., 1997; Egozcue et al., 1997; Guttenbach et al., 1997) and these differences are generally thought to represent technical rather than biological variations. Although differences in the scoring criteria are suspected to be the major sources of variation across laboratories, this question has not been addressed experimentally in an inter-laboratory study.
The purpose of this study was to investigate the sources of technical variation of the mESA assay in the context of genetic toxicology. Vinblastine (VBL), a potent aneugen in mouse somatic (Liang and Satya-Prakash, 1985; Manca et al., 1990; Zijno et al., 1989) and female germ cells (Mailhes et al., 1993; Russo and Pacchierotti, 1988), was selected as the chemical for investigation. All animal handling, VBL treatments and sperm isolations were performed at one laboratory (GSF) by one technician. Epididymal sperm were isolated 22 days after treatment to sample cells exposed during meiosis. Slide preparations and hybridizations were prepared by laboratory-specific protocols for the mESA assay and evaluated by scorers from both laboratories before and after harmonization. This paper describes the results of this inter-laboratory evaluation of the mESA assay.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The two experiments described in this paper each involved a separate dosing of mice with VBL. The first experiment (Figure 1
) involved a single intraperitoneal injection (i.p.) multi-dose VBL exposure (0.5, 1.0 and 2.0 mg/kg) plus concurrent control with preparation, pretreatment, hybridization and scoring of sperm-smear slides conducted entirely at the GSF laboratory. The subsequent reanalyses of this experiment included a subset of sperm-smear slides from the 1.0 mg/kg and control animals that were prepared at the GSF but sent to LLNL where they were hybridized and scored using LLNL procedures. In addition, the original sperm-smears from the 1.0 mg/kg and control animals, which were prepared, pretreated, hybridized and analyzed at the GSF, were also re-analyzed at LLNL. The second experiment (Figure 2) was carried out by exposing mice to 1.0 mg/kg, the dose of VBL that produced the highest disomy rates according to the results of the first experiment. Five treated and five control mice were analyzed, and two mESA slides were prepared per mouse. For each mouse, one slide was hybridized at the GSF and one at LLNL. The GSF scorer then traveled to LLNL for a cross-training (harmonization) exercise outlined in Figure 2a. Two controls and two treated mice were scored during the cross training in which the two scorers discussed their diagnoses of each abnormal cell found. Following the cross-training, the rest of the animals were scored independently by the two scorers (Figure 2b).
|
|
Animals and chemical treatment
Male (102/ElxC3H/El) F1 mice aged 10–14 weeks and weighing 25–29 g were used for both exposure experiments. They were bred in the GSF mouse colony, received food and water ad libitum and were maintained on a 12 h light and dark cycle. Five males were randomly assigned to each treatment and concurrent physiological saline control group and dosed by a single i.p. injection. VBL was obtained from Sigma (Deisenhofen, Germany). The injected volume of dosing solution was 0.1 ml/10 g body weight.
Males were killed 22 days after treatment and sperm were collected from the cauda epididymis. The techniques of preparation and decondensation of epididymal sperm, the preparation of the DNA probes and the hybridization conditions have been previously described (Adler et al., 1996; Lowe et al., 1996; Schmid et al., 1999). They are briefly described below for this study with notations on variations between GSF and LLNL.
Preparation of epididymal sperm
The preparation technique for epididymal sperm, which was performed exclusively at GSF, was based on the procedure described by Lowe et al. (1996) with minor modifications. Both epididymides were dissected from the euthanized animal and placed in a Petri dish. The Caput and Cauda portions were separated and incisions were made into the Cauda portions. Then each Cauda portion was placed individually into Eppendorf cups filled with 300 µl of fetal calf serum. The cups were placed in an Eppendorf incubator at 32°C for 30min to allow the sperm to swim out of the epididymides. Epididymides were then removed from the cups and the sperm suspensions frozen on dry ice and stored at –80°C. Fresh or thawed sperm suspensions (5 µl) were pipetted onto clean dry glass slides. Cells were smeared across the slide and allowed to dry overnight. The slides were stored at –20°C under nitrogen gas until use. Fresh sperm-smear slides from each experiment to be processed at LLNL were shipped on dry ice.
Preparation of the DNA probes
The plasmid DNA used to make the probes for chromosome 8 (clone 4 and 5e; Boyle and Ward, 1992) and the X chromosome (DXWas70; Disteche et al., 1985) were the same for GSF and LLNL. They were transformed in Escherichia coli XL1-blue and DNA isolations were made using the Qiagen Plasmid Maxi Kit (Qiagen, Chatsworth, MD) as described by Lowe et al. (1995). The DNA for the Y chromosome-specific probe used by GSF was prepared at GSF from flow-sorted murine chromosomes by primer-directed DNA amplification using polymerase chain reaction (PCR) with the JUN1-primer and the UW4B primer (Weier et al., 1994). The DNA for the Y probe used by LLNL, on the other hand, was obtained from Breneman et al. (1995).
The probes used by GSF were indirectly-labeled while those used by LLNL were directly-labeled. GSF labeled the chromosome 8 probe with bio-16-dUTP, the Y probe with dig-11-dUTP and the X probe with a combination of bio-16-dUTP and dig-11-dUTP using the Gibco Nick Translation System (Gibco BM, Germany). The probes were detected by streptavidine-CY3 (chromosome 8) and anti-dig-FITC (Y chromosome) or the combination of both (X chromosome). In the first experiment at LLNL, the probe for chromosome 8 was labeled with Cy3-dUTP (Amersham, Arlington Heights, IL) and the probe for the X was labeled with fluorescein-12-dUTP (Boehringer-Mannheim) using a modified nick translation system (Gibco BRL). The painting probe for the Y chromosome was labeled with fluorescein-12-dUTP and rhodamine-6-dUTP (Boehringer, Mannheim) by PCR. Thus, the signal for chromosome 8 was red for both GSF and LLNL stained cells; however, GSF stained cells had a yellow signal for the X chromosome and a green signal for the Y, which was reversed for LLNL protocol. However, in the second experiment, the probe labeling strategy at LLNL was modified to correspond to that used at GSF to avoid confusion during the cross-training and subsequent scoring.
Coding of slides
At GSF, slides were coded prior to hybridization by an individual other than the one performing the pretreatments, hybridizations and scoring. A 60x22 mm cover slip was used, therefore the entire slide was available for the analysis. At LLNL, the slides were coded after pretreatments and hybridizations by an individual who did not participate in the scoring. To avoid possible scoring bias based on cell concentration, special effort was made to select a 22x22 mm area for analysis, which exhibited uniform sperm density to assure that slides made from control and treated mice were indistinguishable to the scorer.
Pretreatment and hybridization
The method for decondensation of sperm was essentially that previously described by Lowe et al. (1996). Slides were incubated in a Coplin jar in 10 mM dithiothreitol (Sigma) for 30 min on ice followed by incubation in 4 mM lithium-3,5-diiodosalicylic acid (LIS, Sigma). At the GSF, the incubation with LIS was for 30 min on ice while that at LLNL was for 60 min at room temperature. At the GSF, slides were subjected to drying both prior to and following decondensation by heating them on a 70°C hot plate for 5 min. Slides at LLNL were air-dried only following decondensation.
Hybridizations were performed according to the procedures of Lowe et al. (1996) with some modifications as described by Adler et al. (1996). Labeled probes were mixed with Master Mix 2.1 (55% formamide, 10% dextran in 1x SSC) and denatured at 78°C for 8 (GSF) or 6 min (LLNL). The sperm slides were denatured in 70% formamide (in 2x SSC, pH 7.0) at 78°C for 5 (GSF) or 6 min (LLNL) and then dehydrated at room temperature in an alcohol series consisting of 2 min each in 70, 90 (GSF) or 85 (LLNL) and 100% ethanol. GSF slides were dried on a slide warmer at 37°C for 3 min, while LLNL slides were air-dried, prior to application of the denatured hybridization mix. After applying the denatured hybridization mix and coverslip, the slides were incubated 24–48 h (GSF) or overnight (LLNL) in a moist chamber at 37°C. Post-hybridzation washings consisted of 50% formamide (2x SSC, pH 7.0) at 45°C (15 min for GSF and 10 min for LLNL) and PN-buffer (30 min at 37°C for GSF, and 10 min at 37°C plus 10 min at room temperature for LLNL). For GSF, probes were detected by immunofluorescence through the application of streptavidin-CY3 (chromosome 8) or anti-dig-FITC (Y chromosome) or the combination of both (X chromosome). Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI). At GSF, 0.1 µg/ml DAPI in phosphate-buffered saline was placed on the slide for 10 min at room temperature before cover-slipping in Vecashield (Vector Labs., CA). At LLNL, DAPI at 0.125 µg/ml was applied to the slide directly in the Vectashield mounting media. Slides were stored at 4°C in the dark.
Scoring
10 000 sperm per animal were scored using a Zeiss Axiopan Fluorescence Photo Microscope (Zeiss, Germany). The microscope was equipped with a triple band-pass filter (set no. 61000, Chroma Technology, Brattleboro, VT) for simultaneous visualization of green (FITC), yellow (FITC + CY3), red (CY3) fluorescence hybridization domains, as well as the blue (DAPI) fluorescence of the sperm nuclei, and with individual filters for each of the fluorochromes (FITC: HQ 480/40, HQ 535/50; CY3: HQ 535/50, HQ 610/75, Chroma Technology, Brattleboro, VT, and DAPI: BP 365, LP 397, Zeiss, Germany). A similar microscope setting was used at LLNL. At the GSF, records of abnormal sperm were made by digitizing the microscopic image using the computer program ISIS3 (MetaSystems, Altlussheim, Germany). At the LLNL, records of all sperm scored were maintained using a special program (CYTOscore) developed at LLNL for the Macintosh computer.
The sperm were assigned to the specific fluorochrome phenotypes as determined by the combination of fluorescence signals in each nucleus: X-8 and Y-8 as normal sperm. Strict scoring criteria were employed for assigning cells to one of five hyperhaploidy classes (X-X-8, Y-Y-8, X-Y-8, X-8-8, Y-8-8), to X-Y-8-8 as diploid sperm resulting from complete failure of the first meiotic division, and to X-X-8-8 or Y-Y-8-8 as autodiploid sperm resulting from failure of the second meiotic division. Cells were scored as having two domains of the same color if both signals were of similar size and intensities, and were separated by a distance of more than half the diameter of one domain. The frequencies of sperm with hypohaploid phenotypes (X-0, Y-0, 0-8) were scored but not included in the statistical analysis.
During the cross-training, in which slides from two controls and two VBL-treated mice were analyzed, both scorers examined each sperm identified by either scorer as having an abnormal phenotype and discussed whether there was any disagreement as to the call and the criteria used to judge that phenotype as abnormal. After the cross-training, slides from three controls and three VBL-treated mice were scored independently by both scorers.
Statistical analysis
If no significant animal to animal variability was present, 
2 test with Yate's correction was used to compare treated versus control frequencies of disomic or diploid sperm. If significant animal to animal variation was present, these frequencies were compared by using a Mann–Whitney U-test (Siegel, 1956).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Table I
shows a comparison of the baseline frequencies of aneuploid sperm in two different stocks of healthy young adult untreated male mice based on the analysis of >600 000 sperm from 55 control mice scored at LLNL or GSF in previous experiments using the three-chromosome mESA assay. A statistical analysis of the 25 control animals scored at LLNL and 30 control animals scored at the GSF showed that the frequency of disomic sperm was significantly (P < 0.05) higher at LLNL (~2-fold). This difference was due mostly to higher frequencies of X-X-8, X-8-8 and Y-8-8 sperm among the LLNL mice. These comparisons show that there were specific and systematic differences in baseline frequencies between two laboratories. It is not known whether these differences were due to the different animal strains used by the two laboratories or to technical factors related to the mESA assay.
|
To remove the genotype and animal handling variables, we conducted an experiment in which treatment and FISH preparations were done at the GSF, while scoring was performed at both laboratories. First, the effects of VBL exposure on chromosome segregation during male meiosis were analyzed using the three-chromosome mESA assay on slides pretreated, hybridized and scored at the GSF (Table II
). The GSF scorer found no significant differences in the treated group versus controls for the sex ratio of sperm, no significant treatment effects on the frequencies of diploid sperm at any dose, and no treatment effect on disomy in the 0.5 mg/kg VBL group. However, compared with the corresponding controls (0.057 and 0.050%), the frequencies of disomic sperm at 1.0 and 2.0 mg/kg of VBL (0.095 and 0.086%, respectively) were significantly increased (P < 0.05) by a factor of <2 (Table II).
|
A subset of slides (air-dried sperm, but not otherwise treated) from the animals treated with 1.0 mg/kg VBL from this first experiment were then sent to LLNL where they were prepared for microscope analyses using the LLNL FISH hybridization procedure with the same combination of chromosome-specific probes used at GSF (Table III
, columns 1 and 2). In disagreement with the findings at the GSF, no significant differences were detected by the LLNL scorer between the VBL-treated animals and the corresponding controls (0.085 and 0.099%, respectively). Because this discordant result may have been due to some differences between either the FISH protocols and/or the scoring criteria used in the two laboratories, another scoring was conducted at LLNL in which the exact same slides scored at GSF for the 1.0 mg/kg VBL exposure plus controls were sent to LLNL to be rescored by the LLNL scorer (Table III, columns 3 and 4). Again, no significant differences were observed by the LLNL scorer in the frequencies of disomic sperm between treated and control mice. Interestingly, the frequencies of disomic sperm determined for exposure to 1.0 mg/kg of VBL at the GSF (0.095%) and at LLNL (0.085 and 0.097%) were very similar. However, the LLNL scorer observed a higher frequency of total disomic sperm (0.099 and 0.116%) in the control animals in both scorings, which was consistent with the other LLNL control values (Table 1
) and was ~2-fold greater than the GSF-control scores (0.059, 0.057 and 0.050). The difference was mostly due to higher scores at LLNL than at the GSF in the categories of X-X-8, X-8-8 and Y-8-8 sperm, also confirming the trend shown in Table I. The affected categories of aneuploid sperm contained two domains of similar color and morphology suggesting that the two scorers differed in their criteria for determining the distance separating the domains. However, the results in Table III show that even when the same animals and slides were used, different answers were obtained by the two laboratories, suggesting that strain difference was not an issue and pointing to the importance of scoring criteria.
|
A second VBL treatment experiment was then conducted at the GSF (Figure 2
) using 1.0 mg/kg VBL. Air-dried sperm smears were prepared at the GSF and one slide from each animal (not otherwise preprocessed) was sent to LLNL. FISH preparations were made at both the GSF and LLNL, but scoring was planned to be conducted at one site (LLNL) by both LLNL and GSF scorers. Before blind scoring began at LLNL, a cross-training (harmonization) exercise was conducted during which both scorers shared with each other their evaluation of each abnormal cell found and discussed their criteria for selection. The cross-training was followed by a blind and fully-independent analysis of the VBL-treated and control animals. As shown in Table IV, there were no significant differences in the frequencies of the disomic sperm among the controls and treated mice during and after the cross training (0.060 and 0.065% for controls; 0.052 and 0.057% for the treated mice). Moreover, the control rates were now consistent with the control rates of the first experiment scored at the GSF (Table II). However, in contrast to the first experiment, no increase of disomic sperm frequencies was found after treatment with 1.0 mg/kg VBL by either the GSF or LLNL scorer, indicating that when these harmonized criteria were applied, no effect of VBL treatment was observed in sperm aneuploidy.
|
Even after harmonization and cross training, there remained a significant (P < 0.01) difference between the two scorers in the prevalences of specific disomies. For example, 50% (21/42) of the disomies diagnosed by the LLNL scorer were X-8-8, while this diagnosis was much less frequent (13/63 = 21%) according to the GSF scorer. Conversely, the most frequent disomy diagnosed by the GSF scorer was X-X-8 (25/63 = 40%), which accounted for only 19% (8/42) of the disomies according to the LLNL scorer (Table IV
). For both scorers, the distribution of disomies was relatively consistent with the corresponding laboratory's previous experience (see Table I). However, one notable difference relative to prior experience was the actual frequency of disomies diagnosed by the LLNL scorer. This frequency (4.79%) is much lower than the previous control frequencies found at LLNL (12.28%; see Table I). For the GSF scorer, the frequencies were in closer agreement (6.98 versus 5.43%; see Tables I and IV). |
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The use of FISH methods to detect numerical abnormalities in sperm of mammals, including human, has gained in popularity because of its relative ease and rapidity of collecting data compared with epidemiological studies of human offspring or animal breeding studies. The results of the present study showed that: (i) differences in scoring criteria can significantly affect the estimates of sperm with numerical aberrations, and (ii) even when similar scoring criteria are employed, 2-fold differences between scorers occur with the mESA. Our study suggests that when small treatment-related differences in the frequency of aneuploidy are found by a scorer (i.e., ~2-fold or less), the slides should be re-evaluated by a second scorer and the entire experiment (treatment to analysis) should be repeated.
The technique to detect aneuploidy in epididymal sperm of mice using multicolor FISH developed by Wyrobek et al. (1995) and Lowe et al. (1996), was used previously by Schmid et al. (1999) to investigate chemically-induced aneuploidy in mouse sperm. The current study was motivated by the need to evaluate the reliability of small (<2-fold) differences in sperm aneuploidy frequencies observed using the mESA assay. In the current study, when the first set of sperm slides from the 1.0 and 2.0 mg/kg groups were scored after multicolor FISH at GSF, the frequencies of disomic sperm showed a small but significant elevation above the concurrent controls (Table II). However, a re-evaluation of the same animals using LLNL hybridization and scoring procedures (Table III) did not confirm the VBL effect. At that point, it was possible that the discordant VBL findings between the two laboratories might be due to FISH protocol or scoring criteria. The role of the FISH protocol in the inter-laboratory difference was removed when the LLNL scorer could not confirm the GSF finding of a VBL effect even when the exact same GSF prepared FISH slides were evaluated. Thus, scoring criteria differences seemed the likely answer for the disparate VBL effects observed in the two laboratories.
In the first VBL experiment, the control frequencies obtained by the GSF and LLNL scorer varied significantly (P < 0.05) by a factor of almost 2 (Tables II and III). A general inter-laboratory difference in control frequencies with the mESA was previously noted by Adler et al. (1996) and continues to be evident (Table 1). Since the difference observed by the GSF scorer between control and VBL-treated mice might have been driven in part by the lower GSF control values, a harmonization of scoring criteria and particularly of the control values was undertaken at one site (LLNL). We conducted a second VBL-treatment experiment (Figure 2) and, immediately after harmonization, had GSF and LLNL scorers independently evaluate VBL slides in a rigorously blind design. The findings of the first VBL experiment, both for the increased frequency of disomic sperm at the GSF (Table 2) and for the high frequency of disomic sperm in the control groups at the LLNL (Table 3), were not confirmed in the second VBL experiment (Table 4). Thus, we conclude that when harmonized scoring criteria were used, in vivo exposure of male mice to 1.0 mg/kg VBL during meiosis did not increase the frequency of aneuploid sperm.
The findings of the present study establish that harmonization of scoring criteria is a critical underpinning for the reliability of the mESA assay for assessing the chemical induction of sperm aneuploidy. At the beginning of our study, both laboratories claimed they used the same cell scoring criteria. Before the cross-training, there were distinct differences in the frequencies of some specific categories: e.g., X-X-8 and X-8-8 sperm. After cross-training, several of these differences remained. Inspection of affected cells under the microscope indicated that the morphology (size and shape) of individual X and 8 domains can vary substantially, ranging from a tight spot, to those with a split domain to those made up of a cluster of small spots connected by filaments. There were differences in how scorers determined whether split domains or clusters of spots represent one or two domains (i.e. normal versus disomy). Strict criteria demanded that disomic sperm contained domains separated by at least half average domain diameter, but the decision required subjective assessments of domain size and separation. This difficulty was amplified when parts of the two clusters were in slightly different focal planes within the nucleus. The practitioners of sperm FISH know that there is no substitute for preparing slides that minimize split domains and cluster domains so that, ideally, all domains are tight spots.
Three additional lessons were learned to improve the reliability of the mESA assay. First, the importance of a thorough blinding of the scorers was reaffirmed, even though this is very difficult in toxicology when treatment reduces cell numbers. A small but highly relevant difference was discovered between the GSF and the LLNL FISH protocols. The LLNL protocol requires searching the slides for a region of `normal or near-normal' sperm density, and applying a small (22x22) coverslip to limit the microscope analysis to that region. The GSF protocol uses a coverslip covering the entire slide. Indeed, three experienced microscopists were able independently to correctly identify four out of five coded slides from treated animals prepared by the GSF protocol based solely on the observation of lower sperm density on slides from treated animals. The same microscopists were unable to distinguish between treated and untreated slides using the LLNL protocol. Thus, we strongly encourage that a third person performs the hybridization of the slides and selects the regions for microscopic analysis in order to minimize possible biases or artifacts due to low cell density.
Second, we noted that the determination of the statistical significance of small differences in the percent of disomic sperm could be affected by unexpected `underdispersion', i.e., less variability among animals within a given group than would be expected to occur by chance, based (for example) on a Poisson model. An example of underdispersion in our study is provided in Table II, in which the distribution of total disomies in the five animals from the solvent control II group (6-7-6-5-5) and the 1.0 mg/kg VBL group (8-9-10-11-11) shows very little variability. When marked underdispersion occurs, the statistical analysis may exaggerate the significance of small differences among groups.
Third, the post-harmonization results for this study (Table IV) showed that after a period of cross-training, in which each scorer double-checked the abnormal cells found by the other, one scorer was still observing ~50% more aneuploid sperm than the other (7.2 for T.E.S. versus 4.9 for C.Sanders). Furthermore, there remained a significant difference in the distribution of specific disomies found by each scorer, with certain disomies showing 3–6-fold differences between scorers (Table IV). This strongly indicates that 
2-fold variations may be unavoidable with the current mESA assay. Meanwhile, a series of chemicals has been tested by the mESA assay and positive results were reported with colchicine, diazepam (Schmid et al., 1999), griseofulvin (Qinghua et al., 1999) and trichlorfon (Sun et al., 2000). However, none of these results exceeded a doubling of the concurrent control frequencies. The findings of our study suggest that differences of a factor of 2 or less between scorers, laboratories, or between treatment groups should be viewed with extreme caution, in spite of statistical significance, until the findings are replicated and/or slides are rescored by a second person.
The present finding of a lack of an effect of VBL on the frequencies of aneuploid sperm as detected by the mESA assay is in contrast with the known aneugenic activity in male bone marrow cells (Liang and Satya-Prakash, 1985; Manca et al., 1990; Zijno et al., 1989) and oocytes (Mailhes et al., 1993; Russo and Pacchierotti, 1988) and adds to the list of contrasting results in male germ cells. VBL treatment caused meiotic delay (Miller and Adler, 1992), increased the frequencies of hyperploid secondary spermatocytes (Miller and Adler, 1992), and induced spindle aberrations in spermatocytes (Gassner and Adler, 1995). However, Liang et al. (1986) found no significant induction of hyperhaploid spermatocytes in mice after treatment of spermatogonial as well as zygotene cells. These differing outcomes may be due to differences in dose, dose regimen, strain of mice used and biological differences related to the endpoint measured.
It is particularly of note that the results in male meiosis are vastly different from the unequivocal positive results obtained in mouse oocytes (Mailhes et al., 1993; Russo and Pacchierotti, 1988). The present data and those obtained with griseofulvin (Qinghua et al., 1999) point to major differences between males and females in their germ-cell responses to spindle inhibitors (Eichenlaub-Ritter et al., 1996). Unlike mitotic and male meiotic cells, oocytes do not contain centrioles (Messinger and Albertini, 1991). Instead, they contain multiple microtubule organizing centers (MTOC) which form ring- or disk-like structures at the spindle poles (Szöllösi et al., 1972). Additionally, the cell cycle checkpoint controlling chromosome alignment in metaphase (Taylor et al., 1998), which arrests the mitotic and male meiotic cycle if the mechanical attachment of each pairs of kinetochores to two oppositely directed spindle fibers is not complete (Rieder et al., 1994; Nicklas, 1997; Rieder and Salmon, 1998), does not operate in mammalian oocytes (Hunt et al., 1995). Whether these differences are the main causes for sensitivity differences for aneuploidy induction by spindle inhibitors between the two sexes remain to be determined.
In conclusion, the present study has identified several technical factors that are critical to the reliability of the mESA assay when small changes are observed among laboratories, scorers and treatment groups. These include harmonizing scoring criteria, rigorously blinding scorers using procedures that normalize cell numbers, evaluating dispersion characteristics of the control and treatment groups and replicating findings in repeated experiments using harmonized scorers. It might also be desirable to expand the sample sizes (number of animals per group) in order to obtain more statistical power. Furthermore, because microscopic scoring is laborious and time-consuming, and automated methods are urgently needed (e.g., flow-cytometric analysis, automated analysis by laser scanning cytometry or computer-controlled microscopy).
|
Acknowledgments |
|---|
We are grateful Christina Sanders who scored the slides at LLNL and to Dr A.Baumgartner who contributed to the many discussions during scoring harmonization and interpretation of the data. We are especially grateful to Dr Ilse-Dore Adler who provided her support for this project including treatment of animals, the inter-laboratory harmonization study and critical review of this manuscript. The studies were supported by the EU contracts EV5V-CT94-0403 and ENV4-CT97-0471. This work was performed in part under the auspices of the US Department of Energy by the Lawrence Livermore National Laboratory under contract no. W-7405-Eng-48 with support by NIEHS-DOE/LLNL Interagency Agreement Y01-ES-10203.
|
Notes |
|---|
4 These authors contributed equally to this work
5 Present address: Department of Psychiatry, California Pacific Medical Center, San Francisco, CA 94120, USA
6 To whom correspondence should be addressed. Tel: +1 925 422 6296; Fax: +1 925 424 3130; Email: wyrobek1{at}llnl.gov
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
-
Adler,I.D., Bishop,J., Lowe,X., Schmid,T.E., Schriever-Schwemmer,G., Xu,W. and Wyrobek,A.J. (1996) Spontaneous rates of sex chromosomal aneuploidies in sperm and offspring of mice: a validation of the detection of aneuploid sperm by fluorescence in situ hybridization. Mutat. Res., 372, 259–268.[Web of Science][Medline]
Baulch,J.E., Lowe,X.R., Bishop,J.B. and Wyrobek,A.J. (1996) Evidence for a parent-of-origin effect on sperm aneuploidy in mice carrying Robertsonian translocations as analyzed by fluorescence in situ hybridization. Mutat. Res., 372, 269–278.[Web of Science][Medline]
Baumgartner,A., Van Hummelen,P., Lowe,X.R., Adler,I.D. and Wyrobek,A.J. (1999) Numerical and structural chromosomal abnormalities detected in human sperm with a combination of multicolor FISH assays. Environ. Mol. Mutagen., 33, 49–58.[Web of Science][Medline]
Boyle,A.L. and Ward,D. (1992) Isolation and initial characterization of a large repeat sequence element specific to mouse chromosome 8. Genomics, 12, 517–525.[Web of Science][Medline]
Breneman,J.W., Swiger,R.R., Ramsey,M.J., Minkler,J.L., Eveleth,G.G., Langlois,R.A. and Tucker,J.D. (1995) The development of painting probes for dual-color and multiple chromosome analysis in the mouse. Cytogenet. Cell Genet., 68, 197- 202.[Web of Science][Medline]
Disteche,C.M., Tantravahi,U., Gandy,S., Eisenhard,M., Adler,D. and Kunkel,L.M. (1985) Isolation and characterization of two repetitive DNA fragments located near the centromere of the mouse X chromosome. Cytogenet. Cell Genet., 39, 262–268.[Web of Science][Medline]
Downie,S.E., Flaherty,S.P. and Matthews,C.D. (1997) Detection of chromosomes and estimation of aneuploidy in human spermatozoa using fluorescence in-situ hybridization. Mol. Hum. Reprod., 3, 585–598.
Egozcue,J., Blanco,J. and Vidal,F. (1997) Chromosome studies in human sperm nuclei using fluorescence in-situ hybridization (FISH). Hum. Reprod. Update, 3, 441–452.
Eichenlaub-Ritter,U., Baart,E., Yin,H. and Betzendahl,I. (1996) Mechanisms of spontaneous and chemically-induced aneuploidy in mammalian oogenesis: basis of sex-specific differences in response to aneugens and the necessity for further tests. Mutat. Res., 372, 279–294.[Web of Science][Medline]
Gassner,P. and Adler,I.D. (1995) Analysis of chemically induced spindle aberrations in male mouse germ cells: comparison of differential and immunofluorescent staining procedures. Mutagenesis, 10, 243–252.
Guttenbach,M., Engel,W. and Schmid,M. (1997) Analysis of structural and numerical chromosome abnormalities in sperm of normal men and carriers of constitutional chromosome aberrations. A review. Hum. Genet., 100, 1–21.[Web of Science][Medline]
Hassold,T.J. (1986) Chromosome abnormalities in human reproductive wastage. Trends Genet., 2, 105–110.
Hunt,P., LeMaire,R., Embury,P., Sheean,L. and Mroz,K. (1995) Analysis of chromosome behavior in intact mammalian oocytes: monitoring the segregation of a univalent chromosome during female meiosis. Hum. Mol. Genet., 4, 2007–2012.
Jacobs,P.A. (1992) The chromosome complement of human gametes. In Milligan,S.R. (ed.) Oxford Reviews of Reproductive Biolology. Oxford University Press, New York, NY, pp. 47–72.
Liang,J.C. and Satya-Prakash,K.L. (1985) Induction of aneuploidy by mitotic arrestants in mouse bone marrow. Mutat. Res., 155, 61–70.[Web of Science][Medline]
Liang,J.C., Sherron,D.A. and Johnston,D. (1986) Lack of correlation between mutagen-induced chromosomal univalency and aneuploidy in mouse spermatocytes. Mutat. Res., 163, 285–297.[Web of Science][Medline]
Lowe,X., Collins,B., Allen,J., Titenko-Holland,N., Breneman,J., van Beek,M., Bishop,J. and Wyrobek,A.J. (1995) Aneuploidies and micronuclei in the germ cells of male mice of advanced age. Mutat. Res., 338, 59–76.[Web of Science][Medline]
Lowe,X., O'Hogan,S., Moore,D.II, Bishop,J. and Wyrobek,A. (1996) Aneuploid epididymal sperm detected in chromosomally normal and Robertsonian translocation-bearing mice using a new three-chromosome FISH method. Chromosoma, 105, 204–210.[Web of Science][Medline]
Mailhes,J.B., Aardema,M.J. and Marchetti,F. (1993) Investigation of aneuploidy induction in mouse oocytes following exposure to VBL-sulfate, pyrimethamine, diethylstilbestrol diphosphate, or chloral hydrate. Environ. Mol. Mutagen., 22, 107–114.[Web of Science][Medline]
Manca,A., Bassani,B., Russo,A. and Pacchierotti,F. (1990) Origin of aneuploidy in relation to disturbances of cell-cycle progression. I. Effects of vinblastine on mouse bone marrow cells. Mutat. Res., 229, 29–36.[Web of Science][Medline]
Messinger,S.M. and Albertini,D.F. (1991) Centrosome and microtubule dynamics during meiotic progression in the mouse oocyte. J. Cell Sci., 100, 289–298.
Miller,B.M. and Adler,I.D. (1992) Aneuploidy induction in mouse spermatocytes. Mutagenesis, 7, 69–76.
Nicklas,R.B. (1997) How cells get the right chromosomes. Science, 275, 632–637.
Qinghua,S., Schmid,T.E. and Adler,I.D. (1999) Griseofulvin-induced aneuploidy and meiotic delay in male mouse germ cells: detected by using conventional cytogenetics and three-color FISH. Mutat. Res., 441, 181–190.[Web of Science][Medline]
Rieder,C.L. and Salmon,E.D. (1998) The vertebrate cell kinetochore and its roles during mitosis. Trends Cell Biol., 8, 310–318.[Web of Science][Medline]
Rieder,C.L., Schultz,A., Cole,R. and Sluder,G. (1994) Anaphase onset in vertebrate somatic cells is controlled by a checkpoint that monitors sister kinetochore attachment to the spindle. J. Cell Biol., 127, 1301–1310.
Robbins,W.A., Baulch,J.E., Moore,D.I., Weier,H.U., Blakey,D. and Wyrobek,A.J. (1995) Three-probe fluorescence in situ hybridization to asses chromosome X, Y, and 8 aneuploidy in sperm of 14 men from two healthy groups: Evidence for a paternal age effect on sperm aneuploidy. Reprod. Fertil. Dev., 7, 799–809.[Medline]
Robbins,W.A., Meistrich,M.L., Moore,D., Hagemeister,F.B., Weier,H.U., Cassel,M.J., Wilson,G., Eskenazi,B. and Wyrobek,A.J. (1997a) Chemotherapy induces transient sex chromosomal and autosomal aneuploidy in human sperm. Nature Genet., 16, 74–78.[Web of Science][Medline]
Robbins,W.A., Segraves,R., Pinkel,D. and Wyrobek,A.J. (1993) Detection of aneuploid human sperm by fluorescence in situ hybridization: evidence for a donor difference in frequency of sperm disomic for chromosomes 1 and Y. Am. J. Hum. Genet., 52, 799–807.[Web of Science][Medline]
Robbins,W.A., Vine,M.F., Young Truong,K. and Everson,R.B. (1997) Use of fluorescence in situ hybridization (FISH) to assess effects of smoking, caffein, and alchool on aneuploidy load in sperm of healthy men. Environ. Mol. Mutagen., 30, 175–183.[Web of Science][Medline]
Rubes,J., Lowe,X., Moore,D.I., Perreault,S., Slott,V., Everson,D., Selevan,S. and Wyrobek,A.J. (1998) Cigarette-smoking lifestyle is associated with increased sperm disomy in teenage men. Fertil. Steril., 70, 715–723.[Web of Science][Medline]
Russo,A. and Pacchierotti,F. (1988) Meiotic arrest and aneuploidy induced by vinblastine in mouse oocytes. Mutat. Res., 202, 215–221.[Web of Science][Medline]
Schmid,T.E., Xu,W. and Adler,I.D. (1999) Detection of aneuploidy by multicolor FISH in mouse sperm after in vivo treatment with acrylamide, colchicine, diazepam or thiabendazole. Mutagenesis, 14, 173–179.
Siegel,S. (1956) Non-Parametric Statistics. McGraw-Hill, New York, NY.
Spriggs,E., Rademaker,A. and Martin,R.H.(1995) Aneuploidy in human sperm: results of two- and three-colour fluorescence in situ hybridization using centromeric probes for chromosomes 1, 12, 15, 18, and Y. Cytogenet. Cell Genet., 71, 47–53.[Web of Science][Medline]
Sun,F.Y., Schmid,T.E., Schmid,E., Baumgartner,A. and Adler,I.D. (2000) Trichlorfon induces spindle disturbances in V79 cells and aneuploidy in male mouse germ cells. Mutagenesis, 17, 17–24.
Szöllösi,D., Calarco,P. and Donahue,R.P. (1972) Absence of centrioles in the first and second meiotic spindles of mouse oocytes. J. Cell Sci., 11, 521–541.
Taylor,S.S., Ha,E. and McKeon F. (1998) The human homologue of Bub3 is required for kinetochore localization of Bub1 and a Mad3/Bub1-related protein kinase. J. Cell. Biol., 142, 1–11.
Weier,H.U., Polikoff,D., Fawcett,J.J., Greulich,K.M., Lee,K.H., Cram,S., Chapman,V.M. and Gray,J.W. (1994) Generation of five high-complexity painting probe libraries from flow-sorted mouse chromosomes. Genomics, 21, 641–644.[Web of Science][Medline]
Wyrobek,A., Lowe,X., Pinkel,D. and Bishop,J. (1995) Aneuploidy in late-step spermatids of mice detected by two-chromosome fluorescence in situ hybridization. Mol. Reprod. Dev., 40, 259–266.[Web of Science][Medline]
Wyrobek,A.J. (1993) Methods and concepts in detecting abnormal reproductive outcomes of paternal origin. Reprod. Toxicol., 7, 3–16.
Zijno,A., Quaggia,S. and Pacchierotti,F. (1989) A cytogenetic approach to evaluate in vivo somatic aneuploidy. Effects of diethylstilboestrol on mouse bone marrow cells. Mutagenesis, 4, 62–66.
Received on July 28, 2000; revised on October 31, 2000; accepted on August 12, 2000.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. J. Wyrobek, T. E. Schmid, and F. Marchetti Relative Susceptibilities of Male Germ Cells to Genetic Defects Induced by Cancer Chemotherapies J Natl Cancer Inst Monographs, March 1, 2005; 2005(34): 31 - 35. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||