Efficient Monitoring of In Vivo Pig-a Gene Mutation and Chromosomal Damage: Summary of 7 Published Studies and Results From 11 New Reference Compounds

Abstract

The ability to effectively monitor gene mutation and micronucleated reticulocyte (MN-RET) frequency in short-term and repeated dosing schedules was investigated using the recently developed flow cytometric Pig-a mutation assay and flow cytometric micronucleus analysis. Eight reference genotoxicants and three presumed nongenotoxic compounds were studied: chlorambucil, melphalan, thiotepa, cyclophosphamide, azathioprine, 2-acetylaminofluorene, hydroxyurea, methyl methanesulfonate, o-anthranilic acid, sulfisoxazole, and sodium chloride. These experiments extend previously published results with seven other chemicals. Male Sprague Dawley rats were treated via gavage for 3 or 28 consecutive days with several dose levels of each chemical up to the maximum tolerated dose. Blood samples were collected at several time points up to day 45 and were analyzed for Pig-a mutation with a dual-labeling method that facilitates mutant cell frequency measurements in both total erythrocytes and the reticulocyte subpopulation. An immunomagnetic separation technique was used to increase the efficiency of scoring mutant cells. Blood samples collected on day 4, and day 29 for the 28-day study, were evaluated for MN-RET frequency. The three nongenotoxicants did not induce Pig-a or MN-RET responses. All genotoxicants except hydroxyurea increased the frequency of Pig-a mutant reticulocytes and erythrocytes. Significant increases in MN-RET frequency were observed for each of the genotoxicants at both time points. Whereas the highest Pig-a responses tended to occur in the 28-day studies, when total dose was greatest, the highest induction of MN-RET was observed in the 3-day studies, when dose per day was greatest. There was no clear relationship between the maximal Pig-a response of a given chemical and its corresponding maximal MN-RET response, despite the fact that both endpoints were determined in the same cell lineage. Taken with other previously published results, these data demonstrate the value of integrating Pig-a and micronucleus endpoints into in vivo toxicology studies, thereby providing information about mutagenesis and chromosomal damage in the same animals from which toxicity, toxicokinetics, and metabolism data are obtained.

Combining multiple genetic toxicology endpoints into a single in vivo study, either one designed specifically for genotoxicity evaluation or integrated into studies designed for the determination of other toxicological endpoints, is attractive because it reduces animal use and enables comprehensive comparative analyses using toxicity, metabolism, and pharmacokinetic information from the same animal (MacGregor et al., 1990; Pfuhler et al., 2009; Rothfuss et al., 2011). This laboratory has developed highly efficient flow cytometric scoring techniques for monitoring both micronucleated reticulocyte (MN-RET) frequency and gene mutation at the phosphatidylinositol glycan-class A (Pig-a) locus, thereby making combination and integration studies practical (Dertinger et al., 2010).

The MN-RET endpoint has been validated over decades of use and is increasingly used in combination and integrated study designs (Recio et al., 2010; Rothfuss et al., 2010). Rodent-based assays for monitoring Pig-a mutations are a more recent development. The Pig-a endpoint is based on the fact that mutation at the X-chromosome Pig-a locus can lead to loss of expression of glycosylphosphatidylinositol (GPI) anchored proteins on the surface of hematopoietic cells (Araten et al., 1999). To date, rodent assays have focused on circulating erythrocytes and use fluorescent antibodies against GPI anchored cell surface proteins such as CD59 or CD24. With this approach, wild-type erythrocytes are intensely fluorescent, whereas mutant erythrocytes are unlabeled. Quantification of nonfluorescent mutant phenotype cells is accomplished via flow cytometric analysis (Bryce et al., 2008; Dobrovolsky et al., 2010; Miura et al., 2008).

Data generated to date indicate that with proper training, the reproducibility and portability of erythrocyte-based Pig-a assays are high (Dertinger et al., 2011c). Furthermore, in an interlaboratory trial that evaluated five prototypical genotoxicants (N-ethyl-N-nitrosourea [ENU], N-methyl-N- nitrosourea, 4-nitroquinoline N-oxide, benzo[a]pyrene, and 7,12-dimethylbenz[a]anthracene), each was found to significantly and reproducibly induce mutant cells in the rat (Bhalli et al., 2011b; Cammerer et al., 2011; Dertinger et al., 2010; Lynch et al., 2011; Shi et al., 2011; Stankowski et al., 2011). Improvements to the scoring method have been made whereby wild-type cells are depleted via immunomagnetic column separation prior to flow cytometric analysis to dramatically increase data acquisition rates and the number of cell equivalents evaluated per sample (Dertinger et al., 2011a). A summary of published results for seven chemicals previously investigated in rat erythrocyte-based Pig-a studies by this laboratory is presented in Table 1.

Table 1.

Previously Published Pig-a Mutation Rat Studies: Chemicals Tested in This Laboratory and in Most Cases At Least One Additional Collaborating Laboratory

Chemical	CAS No.	Vehicle, route of administration	3-Day study doses evaluateda (mg/kg/day)	28-Day study doses evaluateda (mg/kg/day)	Profile	References
ENU	759-73-9	PBS (pH 6), gavage	20*, 40*(Basic)	2.5*, 5*, 10*(Basic)	Ames pos.; in vitro CA pos.; in vivo MN and CA pos.; transgenic pos. (255 pos./67 neg.); carcinogenic	Lambert et al. (2005), Kirkland et al. (2008), Phonethepswath et al. (2010), Dertinger et al. (2010), and Cammerer et al. (2011)
N-Methyl-N-nitrosourea	684-93-5	PBS (pH 6), gavage	15*, 30*(Basic)	2.5*, 5*, 10*(Basic)	Ames pos.; in vitro CA pos.; in vivo MN and CA pos.; transgenic pos. (43 pos./11 neg.); carcinogenic	Lambert et al. (2005), Phonethepswath et al. (2010), Dertinger et al. (2010), and Lynch et al. (2011)
7,12-Dimethylbenzanthracene	57-97-6	Sesame oil, gavage	25*, 50*(Basic)	2.5*, 5*, 10*(Basic)	Ames pos.; in vitro CA and MLA pos.; in vivo MN pos.; transgenic pos. (52 pos./23 neg.); carcinogenic	Lambert et al. (2005), Kirkland et al. (2008), Phonethepswath et al. (2010), Dertinger et al. (2010), and Shi et al. (2011)
4-Nitroquinoline 1-oxide	56-57-5	0.5% Methyl cellulose in water, gavage	12.5*, 25*(Basic)	1.25, 2.5, 5*(Basic)	Ames pos.; in vitro CA and MLA pos.; in vivo MN pos.; transgenic pos. (25 pos./15 neg.); carcinogenic	Lambert et al. (2005), Phonethepswath et al. (2010), Dertinger et al. (2010), and Stankowski (2011)
Benzo(a)pyrene	50-32-8	Sesame oil, gavage	125*, 250*(Basic)	37.5*, 75*, 150*(Basic)	Ames pos.; in vitro CA and MLA pos.; in vivo MN pos.; transgenic pos. (112 pos./28 neg.); carcinogenic	Lambert et al. (2005), Kirkland et al. (2008), Phonethepswath et al. (2010), Dertinger et al. (2010), and Bhalli et al. (2011b)
1,3-Propane sultone	1120-71-4	Water, gavage	20*, 40*, 80*(HT)	12.5*, 25*, 50/37.5*(top dose adjusted to 37.5 on day 15)(HT)	Ames pos.; in vitro CA.; pos.; in vivo MN pos.; carcinogenic	IARC (1974), Abe and Sasaki (1977), Simmon (1979), Weisburger et al. (1981), and Dertinger et al. (2011b)
Pyrene	129-00-0	Sesame oil, gavage	500, 1000(HT)	125, 250, 500(HT)	Mixed in vitro results, typically neg.; in vivo MN neg.; inadequate carcinogenicity data	de Serres and Ashby (1981), IARC (1983), Ashby et al. (1988), and Torous et al. (2012)

Notes. neg. = negative test result, that is, genotoxicity not evident; pos. = positive test result, that is, genotoxicity apparent; Ames = Salmonella reversion assay; in vitro CA = in vitro chromosomal aberration assay; in vitro MLA = mouse lymphoma assay; in vivo CA = in vivo chromosomal aberration assay; in vivo MN = in vivo micronucleus test, typically with bone marrow or blood-derived erythrocytes.

Pig-a scoring protocols that consisted of direct flow cytometric analysis of peripheral blood erythrocytes are accompanied by the word “Basic,” whereas studies that utilized a high throughput method that consisted of flow cytometric analysis of immunomagnetically separated wild-type and mutant cells are indicated by “HT.”

^aThere were some instances when animals were treated at a higher dose level, but these blood samples were not evaluated because maximum tolerated dose was exceeded.

*Dose levels that were associated with statistically significant increases in Pig-a mutant cell frequencies.

Although data from previous investigations of the Pig-a system are promising, additional data are necessary to more fully characterize the merits and limitations of the assay. Among the work needed is a broader examination of different classes of compounds. We therefore studied eight diverse genotoxicants and three non-DNA reactive chemicals to further evaluate the sensitivity and specificity of the Pig-a endpoint in the context of two commonly employed experimental designs. In conjunction with mutant cell frequency determinations, MN-RET frequencies were also collected and evaluated for their ability to respond to these same in vivo exposures.

MATERIALS AND METHODS

Reagents. A list of test agents, along with CAS No., vehicle, dose levels analyzed, and genotoxic profile, is provided in Table 2. Each test agent was purchased from Sigma-Aldrich, St Louis, MO. A positive control chemical, ENU (CAS No. 759-73-9), was also from Sigma-Aldrich. Lympholyte-Mammal cell separation reagent was purchased from CedarLane, Burlington, NC. Anti-PE MicroBeads (i.e., paramagnetic particles), LS Columns, and a QuadroMACS Separator were from Miltenyi Biotec, Bergisch Gladbach, Germany. CountBright Absolute Count Beads and fetal bovine serum were purchased from Invitrogen, Carlsbad, CA. Anticoagulant Solution, Buffered Salt Solution, Nucleic Acid Dye Solution (contains SYTO 13), Anti-CD59-PE, and Anti-CD61-PE were from Prototype High Throughput In Vivo MutaFlow Kits (Litron Laboratories, Rochester, NY). Hank’s Balanced Salt Solution was purchased from MediaTech, Herndon, VA. Reagents used for flow cytometric micronucleus scoring (Anticoagulant Solution, Buffered Salt Solution, Stock Propidium Iodide Solution, Anti-CD71-FITC and Anti-CD61-PE Solutions, Stock RNase Solution, and Malaria Biostandards) were from In Vivo Rat MicroFlow Kits (Litron Laboratories).

Table 2.

Current Pig-a Mutation Studies: Test Articles, Experimental Design Considerations

Chemical	CAS No.	Vehicle, route of administration	3-Day study doses evaluateda (mg/kg/day)	28-Day study doses evaluateda (mg/kg/day)	Profile	References
Chlorambucil	305-03-3	10% EtOH+90% water, gavage	7.5*, 15*	1.25*, 2.5*, 5*	Ames pos.; in vivo CA pos.; in vivo MN and CA pos.; transgenic pos. (22 pos./18 neg.); carcinogen	Lambert et al. (2005)
Melphalan	148-82-3	Sesame oil, gavage	1.25*, 2.5*	0.1875*, 0.375*, 0.75*	Ames pos.; in vivo MN and CA pos.; carcinogen	IARC (1987), Shelby et al. (1989), and Sanderson et al. (1991)
Thiotepa	52-24-4	Water, gavage	3.75, 7.5*, 15*	2.5*, 5*, 10*	Ames pos.; in vitro CA and MN pos.; in vivo CA and MN pos.; scant transgene data; carcinogen	Chen et al. (1998), Lambert et al. (2005), and IARC (1990)
Cyclophosphamide monohydrate	6055-19-2	Water, gavage	7.5*, 15*, 30*	1.25, 2.5*, 5*	Ames pos.; in vitro CA and MLA pos.; in vivo MN and CA pos.; transgenic pos. (24 pos./18 neg.); carcinogen	IARC (1987), Lambert et al. (2005), and Kirkland et al. (2008)
Azathioprine	446-86-6	Sesame oil, gavage	12.5*, 25*, 50*	6.25*, 12.5*	Ames pos.; in vitro CA pos.; in vivo MN and CA pos.; transgenic mixed (3 pos./6 neg.); carcinogen	IARC (1987) and Lambert et al. (2005)
2-Acetylaminofluorene	53-96-3	Sesame oil, gavage	125*, 250*, 500*	37.5*, 75*, 150*	Ames pos.; in vitro CA neg. and MLA pos.; in vivo MN and CA pos.; transgenic pos. (16 pos./12 neg.); carcinogen	Lambert et al. (2005) and Kirkland et al. (2008)
Hydroxyurea	127-07-1	Water, gavage	62.5, 125, 250	31.25, 62.5, 125	Ames neg.; in vitro MN pos.; in vivo MN pos.; scant transgenic data; inadequate carcinogenicity data	IARC (2000) and Lambert et al. (2005)
Methyl methenesulfonate	66-27-3	Water, gavage	22.5*, 45*, 90*	7.5*, 15*, 30*	Ames pos.; in vitro CA and MLA pos.; in vivo MN and CA pos.; transgenic mixed (10 pos./47 neg.); carcinogen	Lambert et al. (2005) and Kirkland et al. (2008)
o-Anthranilic acid	118-92-3	Sesame oil, gavage	500, 1000, 2000	250, 500, 1000	Ames neg.; in vitro CA, MLA and MN pos. at high/toxic conc.; in vivo MN and CA neg.; non-carcinogen	Kirkland et al. (2008)
Sulfisoxazole	127-69-5	Sesame oil, gavage	500, 1000, 2000	250, 500, 1000	Ames neg., MLA pos. at high cytotoxicity; in vivo MN neg.; non-carcinogen	Kirkland et al. (2008)
Sodium chloride	7647-14-5	Water, gavage	500, 1000, 2000	250, 500, 1000	Ames neg.; in vitro CA, and MLA pos. at high conc.; gastric carcinogen	Hirayama (1984), Galloway et al. (1987), and Pottenger (2007)

Notes. neg., negative test result, that is, genotoxicity not evident; pos., positive test result, that is, genotoxicity apparent; Ames, Salmonella reversion assay; in vitro CA, in vitro chromosomal aberration assay; in vitro MLA, mouse lymphoma assay; in vivo CA, in vivo chromosomal aberration assay; in vivo MN, in vivo micronucleus test, typically with bone marrow or blood-derived erythrocytes.

^aThere were some instances when animals were treated at a higher dose level, but these blood samples were not evaluated because maximum tolerated dose was exceeded.

*Dose levels that were associated with statistically significant increases in Pig-a mutant cell frequencies.

Animals, treatments, and blood harvests. Experiments were conducted with the oversight of the University of Rochester’s Institutional Animal Care and Use Committee. Male Sprague Dawley rats were purchased from Charles River Laboratories, Wilmington, MA. Rodents were allowed to acclimate for ~1 week, and their age at the start of treatment was 7 weeks. Water and food were available ad libitum throughout the acclimation and experimental periods. All chemicals were prepared fresh each day of treatment and administered via po gavage at 10ml/kg body weight/day at ~24h intervals.

For the 3-day studies, rats were treated on days 1, 2, and 3 (n = 6 per group). For the 28-day studies, rats were treated on days 1–28 (n = 6 per group). Top-dose levels were based on preliminary dose-range finding experiments. As described in the OECD (2008) Guideline for the Testing of Chemicals, No. 407 OECD, the aim was to find a top-dose level that induced toxic effects but not death or severe morbidity.

There were four test chemicals for which nil Pig-a responses were anticipated: hydroxyurea, o-anthranilic acid, sulfisoxazole, and sodium chloride. For these 3- and 28-day studies, two or three aged-matched rats were administered ENU and served as positive controls. ENU was dissolved in PBS (pH 6.0), and these administrations occurred via po gavage on days 1, 2, and 3 at 20mg/kg/day.

Peripheral blood was collected before the first administration (between days −4 and −1), and again on days 4, 15, 30, and 45 of the 3-day study, and on days 4, 15, 29, and 42 of the 28-day study. (Note there was no “day 0.”) Mutant cell frequencies for the predose blood samples were determined and rats with unusually high RET^CD59− or RBC^CD59− frequencies were deselected from the study based on a one-sided 95% tolerance interval as described previously (Torous et al., 2012). Once sufficient numbers of animals were identified for the study, they were randomly assigned to treatment groups. Based on these criteria, 5.4% of the rats were rejected from the current studies.

Blood was obtained by nicking a lateral tail vein with a surgical blade after animals were warmed briefly under a heat lamp. Approximately 200 µl of free-flowing blood were collected directly into heparinized capillary tubes (Fisher Scientific, CAT No. 22-260-950). For the Pig-a endpoint, 80 µl of each blood sample were transferred to tubes containing 100 µl kit-supplied heparin solution where they remained at room temperature for less than 2h until leukodepletion as described previously (Dertinger et al., 2011a). For the MN-RET endpoint, 30 µl of each whole blood sample were transferred to tubes containing 100 µl kit-supplied heparin solution where they remained at room temperature for less than 2h, after which they were fixed with ultracold methanol as described previously (Torous et al., 2003). For leukocyte counts, 20 µl of whole blood per sample were transferred to empty Eppendorf tubes where they remained on ice for less than 2h.

Leukocyte counts: sample preparation and data acquisition. Leukocyte counts were performed on blood samples collected on day 29 of the 28-day studies. Lymphocyte and neutrophil counts were performed via flow cytometric analyses as described previously (Torous et al., 2012).

Micronucleated reticulocytes: sample preparation and data acquisition. MN-RET frequencies were scored for blood samples collected on day 4 of the 3-day studies, and on days 4 and 29 of the 28-day studies. Quantitation of MN-RET and RET via flow cytometry was performed according to the In Vivo Rat MicroFlow Kit manual and is described in detail elsewhere (Dertinger et al., 2004; Torous et al., 2003). The frequency of MN-RET was based on the acquisition of ~20,000 CD71-positive RETs per blood sample using a BD FACSCalibur flow cytometer running CellQuest Pro v5.2 software. Note that for these micronucleus analyses, RETs were identified based on their characteristic anti-CD71-positive phenotype.

Pig-a mutation: sample preparation and data acquisition. Determinations of Pig-a mutant cell frequencies were performed on blood samples collected predose, and on days 15, 30, and 45 of the 3-day studies, and predose, and on days 15, 29, and 42 of the 28-day studies. Methods for processing blood for Pig-a measurements have been described previously (Dertinger et al., 2011a). Note that for these Pig-a analyses, RETs were identified based on their characteristic SYTO 13-positive phenotype. For several of the most recent studies, a slight variation was used whereby a consistent number of CountBright beads were added at a later point than was described previously, that is, with the final addition of working SYTO 13 solution.

An Instrument Calibration Standard was generated on each day that data acquisition occurred. These samples contained a high prevalence of mutant-mimic cells, and as described previously, provided a means to rationally and consistently define the location of CD59-negative erythrocytes (Phonethepswath et al., 2010). A BD FACSCalibur flow cytometer running CellQuest Pro v5.2 software was used for data acquisition and analysis.

Calculations and statistical analyses. The incidence of RETs and MN-RET is expressed as frequency percent. The formulas used to calculate RBC^CD59− and RET^CD59− frequencies based on data from pre- and postimmunomagnetic separation specimens were described previously (Dertinger et al., 2011a). The formulas used to calculate RBC^CD59− and RET^CD59− frequencies when counting beads were included at a later point are provided as a supplemental file that is available on-line. In all cases, RBC^CD59− and RET^CD59− frequencies are expressed as number per one million total RBCs or RETs, respectively. The prevalence of lymphocytes and neutrophils is provided as number per milliliter. All leukocyte counts, %MN-RET, %RET, mutant cell frequencies, averages, and SE calculations were performed with Excel Office X for Mac (Microsoft, Seattle, WA).

For statistical evaluations, proportions of MN-RET among RETs were transformed in Excel as follows: transformed data = asin[sqrt(proportion of MN-RET)], where “asin” is the arcsine transformation and “sqrt” the square root. RBC^CD59− and RET^CD59− frequencies were log(10) transformed. Because RET^CD59− values of zero were occasionally observed, a 0.1 offset was added to each RET^CD59− value prior to log transformation. Each time point was studied separately, where the effect of treatment on these transformed MN-RET, RBC^CD59−, and RET^CD59− data were compared with concurrent vehicle control using Dunnett’s multiple comparison t-tests in the context of a one-way ANOVA model (JMP, v8.0.1, SAS Institute Inc., Cary, NC). Significance was evaluated at the 5% level using a one-tailed test for increases relative to vehicle control. These same analyses were performed for RET frequencies as well as lymphocyte and neutrophil counts; however, in these cases the tests were two-tailed. Whereas RET frequencies and lymphocyte counts were not transformed, neutrophil counts per milliliter were log(10) transformed in order to satisfy the requirement for a normal distribution.

Historical negative control means and one-sided 95% and 99.99% confidence intervals (upper limit, alpha 0.05) were calculated in JMP for the RBC^CD59− and RET^CD59− endpoints based on predose blood samples from 586 male Sprague Dawley rats (Table 3). Note that these historical control data include animals that were deselected from studies based on their unusually high mutant cell frequencies as described above. For the 11 test chemicals, in order for a statistically significant increase in Pig-a mutant cells to be considered biologically relevant, an RBC^CD59− and/or RET^CD59− frequency needed to exceed the 95% confidence interval (Hayashi et al., 2011). Given the low number of positive control rats used in these studies, significant ENU-induced Pig-a responses were indicated by average RBC^CD59− or RET^CD59− frequency that exceeded the 99.99% confidence interval.

Table 3.

Historical Negative Control Data—Prior to Deselection of Outliers

Endpoint	n	Mean No. (× 10–6)	SD	SEM	One-sided (upper) 95% CI	One-sided (upper) 99.99% CI
Mutant erythrocytes (RBCCD59−)	586	1.19	3.88	0.16	1.45	1.79
Mutant reticulocytes (RETCD59−)	586	1.06	4.07	0.17	1.34	1.69

The correspondence between the maximal MN-RET and Pig-a responses was compared for each of the eight genotoxicants using JMP’s linear regression platform. Two other genotoxicants that were previously reported and that also benefited from the immunomagnetic separation method, 1,3-propane sultone and benzo(a)pyrene, were also included in this analysis. Specifically, each chemical’s highest mean MN-RET frequency (from either the 3- or 28-day study, whichever was greater) was plotted against its corresponding highest RBC^CD59− frequency. JMP’s linear regression platform was used to calculate a best-fit line, an r ² value, and a p value (the latter corresponds to the likelihood that the linear fit describes the data significantly better than a mean fit line; alpha 0.05).

RESULTS

Chlorambucil

The intended top dose of 30mg/kg/day in the 3-day treatment protocol exceeded the maximum tolerated dose, with excessive weight loss and morbidity apparent on day 3. These animals were removed from the study at this time, and 15mg/kg/day became the highest treatment. All groups exposed to chlorambucil exhibited reduced mean weight gain (Table 4). At 15mg/kg/day, %RET values were elevated on day 15 (Fig. 1A), likely due to compensation for significant repression of erythropoiesis during an earlier period. This is supported by the observation that 24h after the last administration, mean %CD71-positive RETs were 6.51% for vehicle-control rats and 0.07% for the 15mg/kg/day animals (Fig. 12A).

Table 4.

Effect of Chemical Treatment on Body Weight and Leukocyte Counts

Chemical and dose (mg/kg/day)	3-day study	28-day study
	Mean weight gain, days 1–4 (g)	Mean weight gain, days 1–29 (g) (% Control)	Lymphocytes/ml (% Control)	Neutrophils/ml (% Control)
Chlor, 0	28.3	200.2 (100%)	7.64 × 106 (100%)	1.28 × 106 (100%)
Chlor, 7.5	−11.5	—	—	—
Chlor, 15	−44.6	—	—	—
Chlor, 30	R3	—	—	—
Chlor, 1.25	—	184.0 (91.9%)	3.33 × 106 (44%)*	916 × 105 (72%)
Chlor, 2.5	—	170.0 (84.9%)	1.19 × 105 (16%)*	5.40 × 105 (43%)*
Chlor, 5	—	91.0 (45.5%)	9.74 × 104 (1%)*	3.97 × 105 (31%)*
Mel, 0	28.2	182.0 (100%)	6.39 × 106 (100%)	1.18 × 106 (100%)
Mel, 1.25	11.3	—	—	—
Mel, 2.5	10.6	—	—	—
Mel, 5	−9.8, R9	—	—	—
Mel, 0.1875	—	186.8 (102.7%)	4.22 × 106 (66%)*	9.23 × 105 (78%)
Mel, 0.375	—	183.4 (100.8%)	1.73 × 106 (27%)*	5.15 × 105 (43%)*
Mel, 0.75	—	126.8 (69.7%)	3.94 × 105 (6%)*	3.30 × 105 (28%)*
Thio, 0	26.8	195.3 (100%)	8.21 × 105 (100%)	1.63 × 106 (100%)
Thio, 3.75	16.0	—	—	—
Thio, 7.5	2.0	—	—	—
Thio, 15	−11.5	—	—	—
Thio, 2.5	—	146.2 (74.8%)	2.072 × 106 (25%)*	5.55 × 105 (34%)*
Thio, 5	—	138.0 (70.6%)	1.21 × 106 (15%)*	7.85 × 105 (48%)*
Thio, 10	—	87.3 (44.7%)	5.21 × 105 (6%)*	3.10 × 105 (19%)*
CP, 0	32.8	171.2 (100%)	7.48 × 106 (100%)	1.89 × 106 (100%)
CP, 7.5	28.5	—	—	—
CP, 15	17.3	—	—	—
CP, 30	9.7	—	—	—
CP, 2.5	—	172.7 (100.9%)	4.63 × 106 (62%)*	1.45 × 106 (77%)
CP, 5.0	—	139.2 (81.3%)	1.44 × 106 (19%)*	1.16 × 106 (62%)
Aza, 0	27.7	205.5 (100%)	9.30 × 106(100%)	9.92 × 106 (100%)
Aza, 12.5	13.7	—	—	—
Aza, 25	20.2	—	—	—
Aza, 50	14.0	—	—	—
Aza, 6.25	—	173.0 (84.2%)	5.52 × 106 (59%)*	485,042 (49%)*
Aza, 12.5	—	152.0 (74.0%)	5.43 × 106 (58%)*	6.59 × 106 (66%)
Aza, 25	—	R22	R22	R22
2AAF, 0	25.8	186.2 (100%)	5.77 × 106 (100%)	2.22 × 106 (100%)
2AAF, 125	−17.7	—	—	—
2AAF, 250	−15.7	—	—	—
2AAF, 500	−25.2	—	—	—
2AAF, 37.5	—	−17.8 (−9.6%)	6.46 × 106 (112%)	1.80 × 106 (81%)
2AAF, 75	—	−21.0 (−11.3%)	6.27 × 106 (109%)	1.62 × 106 (73%)
2AAF, 150	—	6.2 (3.3%)	5.14 × 106 (89%)	1.38 × 106 (62%)
HU, 0	32.3	204.3 (100%)	6.23 × 106 (100%)	7.46 × 106 (100%)
HU, 62.5	28.8	—	—	—
HU, 125	31.8	—	—	—
HU, 250	22.5	—	—	—
HU, 31.25	—	190.0 (93%)	5.75 × 106 (92%)	1.09 × 106 (147%)
HU, 62.5	—	234.4 (114.7%)	4.38 × 106 (70%)	5.52 × 105 (74%)
HU, 125	—	167.4 (81.9%)	3.72 × 106 (60%)	3.74 × 105 (50%)
MMS, 0	26.8	172.6 (100%)	5.11 × 106 (100%)	1.05 × 106 (100%)
MMS, 22.5	13.2	—	—	—
MMS, 45	0.4	—	—	—
MMS, 90	−26.0	—	—	—
MMS, 7.5	—	171.5 (99.4%)	3.76 × 106 (74%)	7.42 × 105 (71%)
MMS, 15	—	169.6 (98.3%)	2.88 × 106 (56%)	1.48 × 106 (141%)
MMS, 30	—	125.6 (72.8%)	3.32 × 106 (65%)	1.20 × 106 (115%)
Anth, 0	28.8	171.2 (100%)	9.29 × 106 (100%)	1.04 × 106 (100%)
Anth, 500	22.8	—	—	—
Anth, 1000	21.0	—	—	—
Anth, 2000	4.5	—	—	—
Anth, 250	—	201.2 (117.5%)	8.21 × 106 (88%)	1.04 × 106 (100%)
Anth, 500	—	260.8 (152.4%)	7.98 × 106 (86%)	1.21 × 106 (117%)
Anth, 1000	—	186.3 (108.9%)	9.38 × 106 (101%)	9.74 × 105 (94%)
Sulf, 0	26.7	206.8 (100%)	9.51 × 106 (100%)	1.02 × 106 (100%)
Sulf, 500	26.5	—	—	—
Sulf, 1000	22.7	—	—	—
Sulf, 2000	24.2	—	—	—
Sulf, 250	—	190.2 (91/9%)	8.74 × 106 (92%)	8.74 × 105 (85%)
Sulf, 500	—	208.8 (101.0%)	7.04 × 106 (74%)	1.17 × 106 (114%)
Sulf, 1000	—	173.0 (83.6%)	1.01 × 107 (107%)	1.07 × 106 (104%)
NaCl, 0	25.5	179.0 (100%)	3.25 × 106 (100%)	8.86 × 105 (100%)
NaCl, 500	25.7	—	—	—
NaCl, 1000	32.4	—	—	—
NaCl, 2000	24.8	—	—	—
NaCl, 250	—	169.0 (94.4%)	3.01 × 106 (93%)	6.99 × 105 (79%)
NaCl, 500	—	211.8 (118.3%)	3.35 × 106 (103%)	6.12 × 105 (69%)
NaCl, 1000	—	190.6 (106.5%)	2.86 × 106 (88%)	6.63 × 105 (75%)

Notes. Chlor, chlorambucil; Mel, melphalan; Thio, thiotepa; CP, cyclophosphamide; Aza, azathioprine; 2AAF, 2-acetylaminofluorene; HU, hydroxyurea; MMS, methyl methanesulfonate; Anth, o-anthranilic acid; Sulf, sulfisoxazole; NaCl, sodium chloride.

R3 = Each of six rats in the highest treatment group removed from study on day 3 due to morbidity; R9 = Each of six rats in highest treatment group removed from study on day 9 due to morbidity; R22 = Three of six rats in highest treatment group removed from study between days 22 and 29 due mortality; treatment group discontinued following day 15 Pig-a measurements.

*Leukocyte counts that were significantly different than vehicle control (two-tailed Dunnett’s test, p < 0.05).

FIG. 1.

Pig-a assay results for chlorambucil experiments, 3-day studies (panels A–C) and 28-day studies (panels D–F). Left-most graphs show mean percentage of reticulocytes (%RETs) versus time; middle graphs show mean mutant reticulocyte (RET^CD59−) frequencies versus time; and right-most graphs show mean mutant erythrocyte (RBC^CD59−) frequencies versus time. All error bars are SEM. Asterisks indicate significance compared with same-day vehicle-control values (i.e., one-sided Dunnett’s t-test, p < 0.05 and mean > historical control 95% confidence interval, upper limit).

FIG. 12.

Mean percentage of blood micronucleated reticulocytes (%MN-RET, Y-axis) and mean percentage of CD71-positive reticulocytes (%RET, YY-axis) are graphed for 3-day and 28-day studies. Error bars are SEM, and asterisks indicate statistical significance compared with vehicle control (one-sided Dunnett’s t-test, p < 0.05).

RET^CD59− frequencies rose rapidly following chlorambucil treatment (Fig. 1B). Maximal values were observed on day 30, when the high-dose group exhibited a mean RET^CD59− value (± one SEM) of 27.4±9.5 × 10^–6, a 34-fold increase over vehicle controls. Elevated RET^CD59− values declined with additional time, yet remained significantly higher than vehicle controls until the last time point studied. RBC^CD59− increased more slowly than RET^CD59−, with maximal values occurring on day 45 (Fig. 1C). At this time, a mean RBC^CD59− value of 11.3±2.3 × 10^–6 was observed for the high-dose group.

Treatment with chlorambucil for 28 consecutive days caused reduced weight gain (Table 4). The two highest dose levels also reduced %RET values at early time point(s) (Fig. 1D). A rebound effect was evident at the high-dose level, with a mean %RET higher than controls on day 42. Neutrophil and lymphocyte counts were reduced in a dose-dependent manner (Table 4).

Protracted dosing with chlorambucil induced significant increases in RET^CD59− frequency on day 15 (Fig. 1E). RET^CD59− frequencies continued to rise, generally reaching maximal values at the last time point studied. As expected, chlorambucil-induced RBC^CD59− frequencies increased more slowly than RET^CD59− and were maximally expressed in the high-dose group (21.1±2.2 × 10^–6) at the day 42 time point (Fig. 1F).

Chlorambucil significantly increased %MN-RET in both study designs, indicating chromosomal damage to hematopoietic cells (Fig. 12A). The greatest response occurred in the 3-day study, where the 7.5mg/kg/day group exhibited a mean MN-RET value of 2.27±0.19%.

Melphalan

Treatment with melphalan for 3 consecutive days led to reduced weight gain (Table 4). The intended top-dose of 5mg/kg/day exceeded the maximum tolerated dose, with excessive weight loss and morbidity apparent on day 9. These animals were removed from the study at this time. Thereafter 2.5mg/kg/day was the highest dose. At this dose, day -15, %RET were markedly elevated over vehicle controls (Fig. 2A). This was likely compensation for significant repression of erythropoiesis during an earlier period, which was evident 24h after the last administration when mean %CD71-positive RET were 8.27% for vehicle-control rats and 1.3% for the 2.5mg/kg/day group (Fig. 12B).

FIG. 2.

Pig-a assay results for melphalan experiments, same format as Figure 1.

RET^CD59− frequencies rose rapidly following melphalan treatment (Fig. 2B). Maximal values were observed on day 15, when the low-dose group exhibited a mean RET^CD59− value of 12.1±1.8 × 10^–6. Only the high-dose group maintained statistical significance beyond day 15. RBC^CD59− frequencies were also elevated in melphalan-treated animals, showing little change between days 15 and 45 (Fig. 2C). The maximal mean RBC^CD59− value was observed at the terminal time point in the high-dose group: 6.3±2.0 × 10^–6, that is, a 9-fold increase over vehicle control.

Treatment with melphalan for 28 consecutive days reduced weight gain in exposed animals (Table 4). The two highest dose levels modestly reduced mean %RET values, but this effect did not attain statistical significance (Fig. 2D). A rebound effect was evident at the high-dose level, as %RET were markedly higher than controls on day 42. Both lymphocyte and neutrophil counts were reduced in a dose-dependent manner, with the greatest reductions observed in lymphocytes (Table 4).

Melphalan treatment caused significant increases in RET^CD59− frequency at each dose level on days 15–42 (Fig. 2E). The highest mean RET^CD59− value, 33.4±13.9 × 10^–6 for the 0.75mg/kg/day dose group, was evident at the last time point studied. RBC^CD59− frequencies exhibited dose-related increases and were maximally expressed in the high-dose group at the last time point (23.2±10.1 × 10^–6; Fig. 2F).

As shown in Figure 12B, melphalan also induced MN-RET in both studies. As reflected by CD71-positive RET frequencies, bone marrow toxicity was appreciable. The greatest micronucleus responses were evident in the 3-day study, where a mean MN-RET frequency of 3.37±0.55% was observed for the 2.5mg/kg/day group.

Thiotepa

Exposure to thiotepa for 3 consecutive days resulted in reduced weight gain (Table 4). The high-dose group exhibited elevated %RET values relative to vehicle controls on day 15 (Fig. 3A). This was likely compensation for appreciable bone marrow toxicity during an earlier period. This is supported by the observation that 24h after the last administration mean %CD71-positive RET were 5.74% for vehicle-control rats and 0.26% for the high-dose group (Fig. 12C).

FIG. 3.

Pig-a assay results for thiotepa experiments, same format as Figure 1.

RET^CD59− frequencies increased rapidly following thio tepa exposure (Fig. 3B), with maximal values occurring on day 15. The highest mutant cell frequency was associated with the 7.5mg/kg/day group, where a mean RET^CD59− value of 16.9±8.3 × 10^–6 was observed. RBC^CD59− also showed a dose-related increase, with the greatest frequencies evident on day 30—14.2±2.7 × 10^–6 for the high-dose group (Fig. 3C).

Treatment with thiotepa for 28 consecutive days reduced weight gain in exposed animals (Table 4). When %RET were measured during Pig-a analyses, thiotepa treatment showed a modest effect at day 42 only (Fig. 3D). However, day 4 CD71-based RET measurements obtained from micronucleus analyses revealed significant dose-dependent reductions (Fig. 12C). Both lymphocyte and neutrophil counts were markedly reduced by thiotepa exposure (Table 4).

With protracted thiotepa administration, RET^CD59− frequencies increased in a dose-dependent manner (Fig. 3E). Maximal or near maximal values were observed on day 29, with little change at the terminal time point. The highest value occurred on day 42, when the top-dose group exhibited a mean RET^CD59− frequency of 71.1±36.7 × 10^–6. RBC^CD59− frequencies were highly elevated in treated rats, with significant increases evident for each exposure level (Fig. 3F). The maximal mean RBC^CD59− value of 68.5±22.3 × 10^–6 was observed in the 10mg/kg/day group on day 42.

As shown in Figure 12C, thiotepa also caused increased MN-RET frequencies in both studies. The greatest micronucleus induction was observed in the 3-day study, where a mean MN-RET frequency of 3.49±0.41% was observed for the 15mg/kg/day group.

Cyclophosphamide

Treatment with cyclophosphamide for 3 consecutive days led to reduced weight gain (Table 4). Mean %RET were significantly affected by cyclophosphamide treatment, with a markedly elevated value evident in the top-dose group on day 15 (Fig. 4A). Stimulated erythropoiesis likely explains these observations, as RET frequencies acquired in conjunction with day 4 micronucleus analyses were markedly reduced: 5.20% for vehicle-control rats and 0.08% for the 30mg/kg/day group (Fig. 12D).

FIG. 4.

Pig-a assay results for cyclophosphamide experiments, same format as Figure 1.

Following exposure to cyclophosphamide, dose-related increases in RET^CD59− frequency were observed on day 15 (Fig. 4B). At later time points, a statistically significant response was maintained in the high-dose group, where the greatest mean RET^CD59− frequency, 12.0±4.6 × 10^–6, was observed at the last time point. In general, the cyclophosphamide responses were rather variable within treatment groups. In fact, one low-dose group rat was so exceptional it was excluded from the graph and all statistical analyses, as its RET^CD59− frequency (and RBC^CD59− frequency to a lesser extent) was so highly elevated relative to all the other animals in the study (i.e., 120.9 and 126.1 × 10^–6 on days 30 and 45, respectively). Mean RBC^CD59− frequencies also showed dose-dependent increases with cyclophosphamide-treatment (Fig. 4C). The maximal response was observed at the termination of the study, when the top-dose group’s mean RBC^CD59− value was 6.3±1.9 × 10^–6, that is, a 9-fold increase over vehicle control.

Treatment with cyclophosphamide for 28 consecutive days resulted in reduced weight gain in rats exposed to 5mg/kg/day (Table 4). %RET was only slightly affected by the high dose on day 42 (Fig. 4D). Whereas lymphocyte counts were reduced by cyclophosphamide exposure, neutrophils were not significantly affected (Table 4).

Cyclophosphamide treatment led to increased RET^CD59− frequencies on days 29 and 42 (Fig. 4E). The maximal RET^CD59− response was observed in the high-dose group on day 29, averaging 5.1±1.7 × 10^–6. Cyclophosphamide-induced RBC^CD59− frequencies increased over the entire study, reaching a maximal mean value of 3.2±0.6 × 10^–6, that is, a 4.6-fold increase over vehicle control (Fig. 4F).

As shown in Figure 12D, mean %RET values indicate that erythropoiesis function was not appreciably affected in the 28-day study, whereas the 3-day study showed evidence of marked bone marrow toxicity. In fact the top dose of 30mg/kg/day provided too few RETs on day 4 to reliably score MN. Cyclophosphamide was observed to significantly increase %MN-RET in both studies. Maximal MN-RET frequencies were observed in the 3-day study’s highest analyzable day 4 exposure group, 15mg/kg/day, and averaged 2.44±0.23%.

Azathioprine

Treatment with azathioprine for 3 consecutive days caused modest reductions to weight gain in exposed rats (Table 4). Whereas %RET measurements associated with Pig-a analyses on days 15–45 did not show evidence of appreciable bone marrow toxicity (Fig. 5A), this endpoint exhibited considerable dose-dependent reductions when measured in conjunction with micronuclei, 24h after the last administration at study day 4 (Fig. 12E).

FIG. 5.

Pig-a assay results for azathioprine experiments, same format as Figure 1.

Dose-related increases in RET^CD59− frequency were evident on day 15 (Fig. 5B). Later time points did not show consistent and statistically significant dose-related increases. The greatest mean RET^CD59− frequency was observed on day 15, where the high-dose group exhibited a mean value of 3.4±1.7 × 10^–6. Mean RBC^CD59− frequencies showed a dose-dependent increase (Fig. 5C). Although the mid-dose group exhibited statistically significant increases in mean RBC^CD59− on days 15 and 30, this was an instance when the mean values did not exceed the historical control’s upper limit mean 95% confident interval (1.45 × 10^–6) and therefore were not deemed biologically significant responses. The maximal RBC^CD59− response was observed at the termination of the study, when the top-dose group’s mean value was 1.7±0.3 × 10^–6, versus 0.9±0.3 × 10^–6 for controls.

Treatment with azathioprine for 28 consecutive days led to reduced weight gain (Table 4). Rats treated with 25mg/kg/day were removed from the study over the course of days 22–29 due to morbidity, so beyond the day 15 time point, 12.5mg/kg/day was the high-dose group. When %RET were measured in conjunction with Pig-a analyses, azathioprine treatment showed a modest effect at day 42 (Fig. 5D). However, day 4 CD71-positive RET measurements performed in conjunction with micronucleus analyses revealed marked dose-dependent reductions (Fig. 12E). Both lymphocyte and neutrophil counts were reduced by azathioprine exposure (Table 4).

Protracted azathioprine treatment increased RET^CD59− frequencies as early as day 15 (Fig. 5E). The day 29 RET^CD59− mean frequencies were elevated, but so variable that statistical significance was not achieved. Less variable RET^CD59− frequencies were observed on day 42, when a maximal mean value of 4.3±1.6 × 10^–6 was observed. Azathioprine-induced RBC^CD59− frequencies increased over the entire time frame studied, reaching a maximal mean value of 4.2±1.9 × 10^–6, an 8.4-fold increase over vehicle control (Fig. 5F).

As shown in Figure 12E, azathioprine treatment significantly induced MN-RET in both studies. The greatest response was observed in the 28-day study, where the high-dose group’s day 4 mean value averaged 1.34±0.19% compared with 0.10±0.01% for controls.

2-Acetylaminofluorene

Exposure to 2-acetylaminofluorene for 3 consecutive days caused appreciable weight loss (Table 4). Each of the treatment groups showed markedly elevated %RET on day 15 relative to controls (Fig. 6A). This is in contrast to %RET values obtained 24h after the last administration, made in conjunction with micronucleus analyses, which showed marked reductions (Fig. 12F).

FIG. 6.

Pig-a assay results for 2-acetylaminofluorene (2-AAF) experiments, same format as Figure 1.

RET^CD59− frequencies increased rapidly following 2-acetylaminofluorene exposure (Fig. 6B), with maximal values occurring on day 15. The greatest responses were observed for the mid-dose group, where rather variable frequencies were obtained. Interestingly, no effect was evident for the high-dose group at any time point studied. RBC^CD59− were also increased with 2-acetylaminofluorene exposure. In agreement with the RET^CD59− endpoint, the responses were quite variable and the highest values were evident in the mid-dose group, where a maximal mean RBC^CD59− value of 10.7±8.1 × 10^–6 was observed (Fig. 6C).

Treatment with 2-acetylaminofluorene for 28 consecutive days reduced weight gain in exposed animals (Table 4). Interestingly, the effect was not as strong in the high-dose group (150mg/kg/day). Whereas the high-dose group was associated with increased %RET on day 15, the other dose levels caused significant reductions at this time point (Fig. 6D). 2-Acetylaminofluorene did not significantly affect lymphocyte or neutrophil counts (Table 4).

RET^CD59− frequencies increased markedly with protracted 2-acetylaminofluorene administration (Fig. 6E). Each dose group was significantly elevated over controls, but again, the responses were not dose related. Maximal values were observed on day 29, when the mid-dose group exhibited a mean RET^CD59− frequency of 57.6±7.3 × 10^–6. RBC^CD59− frequencies were highly elevated in treated rats, where significant increases were observed for each exposure level (Fig. 6F). The maximal mean RBC^CD59− value of 50.2±6.0 × 10^–6 was observed in the 75mg/kg/day group on day 42.

As shown in Figure 12F, 2-acetylaminofluorene increased MN-RET frequencies in both studies. Interestingly, micronucleus induction was greatest in the 28-day study’s day 29 time point, and the frequencies were markedly higher than the same study’s day 4 values. As with the Pig-a endpoint, these responses were not dose related, but rather were saturated at the lowest dose tested.

Hydroxyurea

Treatment with the highest dose level of hydroxyurea for 3 consecutive days modestly affected weight gain (Table 4). Although %RET measurements made in conjunction with Pig-a analyses on days 15–30 showed evidence of stimulated erythropoeisis (Fig. 7A), this endpoint exhibited considerable dose-dependent reductions when measured during micronucleus analyses, that is, 24h after the last administration (Fig. 12G).

FIG. 7.

Pig-a assay results for hydroxyurea experiments, same format as Figure 1. Note that in the case of the positive control, an asterisk indicates the mean mutant cell frequency exceeds the historical control 99.99% confidence interval, upper limit.

RET^CD59− frequencies were not affected by 3 days of hydroxyurea treatment (Fig. 7B). On day 15, one vehicle-control rat exhibited an unusually high RET^CD59− value. Whether this animal was included or excluded from the statistical analyses, no effect on RET^CD59− frequency was observed. Similarly, hydroxyurea treatment had no effect on RBC^CD59− frequencies (Fig. 7C).

Treatment with hydroxyurea for 28 consecutive days only modestly affected weight gain in the high-dose group (Table 4). When %RET were measured during Pig-a analyses, hydroxyurea treatment showed a modest effect in the top-dose group over several time points (Fig. 7D). Day 4 RET measurements performed in conjunction with micronucleus analyses exhibited marked dose-dependent reductions (Fig. 12G). Reductions to mean lymphocyte and neutrophil counts by hydroxyurea did not attain statistical significance (Table 4).

Protracted hydroxyurea treatment did not significantly increase RET^CD59− values (Fig. 7E). Similarly, RBC^CD59− frequencies were consistently low and stable for each treatment group over the duration of the study (Fig. 7F).

As shown in Figure 12G, mean %RET values indicate that erythropoiesis function was markedly affected in both the 3-day and 28-day studies. In fact the top dose of 250mg/kg/day resulted in too few RETs on day 4 for reliable MN-RET scoring. Although hydroxyurea did not affect mutant cell frequencies, it significantly increased MN-RET values in both studies. Maximal MN-RET frequencies were observed on day 29 in the 125mg/kg/day group, which averaged 0.66±0.06%.

Methyl Methanesulfonate

Treatment with methyl methanesulfonate for 3 consecutive days caused dose-dependent reductions in weight gain (Table 4). On day 15, %RET for each treatment group were elevated over vehicle controls (Fig. 8A). This was likely compensation for significant repression of erythropoiesis during an earlier period, for instance day 4 mean CD71-positive RET frequencies were 4.6% for vehicle-control rats and 0.10% for the high-dose methyl methanesulfonate group (Fig. 12H).

FIG. 8.

Pig-a assay results for methyl methanesulfonate (MMS) experiments, same format as Figure 7.

RET^CD59− frequencies were elevated following methyl methanesulfonate treatment (Fig. 8B). Maximal or near maximal values occurred on day 15. Beyond day 15, only the top-dose group exhibited statistically significant increases in RET^CD59−. RBC^CD59− frequencies were also elevated in methyl methanesulfonate-treated animals. In this case, induced frequencies rose throughout the study period, with each dose group exhibiting statistically significant increases on day 45 (Fig. 8C). The greatest RBC^CD59− value was observed for the highest exposure group, 5.7±0.7 × 10^–6.

Treatment with methyl methanesulfonate for 28 consecutive days reduced weight gain in a dose-dependent manner (Table 4). The highest dose level modestly increased %RET values on days 29 and 42 (Fig. 8D). On the other hand, CD71-positive RET values collected on day 4 showed marked reductions to %RET (Fig. 12H). Leukocyte counts were not significantly affected by methyl methanesulfonate treatment (Table 4).

Methyl methanesulfonate treatment for 28 days caused significant increases in RET^CD59− by day 15, but were not maximal until day 29 (Fig. 8E). At this time, the high-dose group exhibited a mean RET^CD59− value of 21±10.3 × 10^–6, a 53-fold increase over concurrent controls. RBC^CD59− frequencies also exhibited dose-related increases and were maximally expressed at the last time point (Fig. 8F). The greatest RBC^CD59− frequency for the top dose studied was 19.2±11.5 × 10^–6.

As shown in Figure 12H, methyl methanesulfonate also induced MN-RET in both studies. Bone marrow toxicity was appreciable as reflected by CD71-positive RET frequencies. In fact the high dose used in the acute study provided too few RETs to reliable score micronuclei. The greatest micronucleus response was observed in the 3-day study, where the 45mg/kg/day dose group averaged 1.43±0.10% MN-RET.

o-Anthranilic acid, Sulfisoxazole, and Sodium Chloride

Weight gain data associated with 3 days of treatment with o-anthranilic acid, sulfisoxazole, and sodium chloride are provided in Table 4. Only the high dose of o-anthranilic acid was observed to affect this parameter. %RET (Figs. 9A, 10A, and 11A) was not affected by any treatment at any time point, with one exception. The high-dose sodium chloride treatment group (2000mg/kg/day) showed modestly elevated %RET frequencies on days 15 and 30.

FIG. 9.

Pig-a assay results for o-anthranilic acid experiments, same format as Figure 7.

FIG. 10.

Pig-a assay results for sulfisoxazole experiments, same format as Figure 7.

FIG. 11.

Pig-a assay results for sodium chloride, same format as Figure 7 .

Treatment of rats for 3 days with o-anthranilic acid, sulfisoxazole, or sodium chloride had no statistically significant effect on RET^CD59− values (Figs. 9B, 10B, and 11B). Similarly, RBC^CD59− frequencies remained stable and no statistically significant effects were observed throughout the study periods (Figs. 9C, 10C, and 11C).

Weight gain data associated with 28 days of treatment with o-anthranilic acid, sulfisoxazole, and sodium chloride are provided in Table 4. Only the high dose of sulfisoxazole modestly affected this parameter. %RET (Figs. 9D, 10D, and 11D) was not affected by any treatment at any time point, with one exception. The mid-dose o-anthranilic acid treatment group showed slightly elevated %RET frequencies on day 15.

Treatment of rats for 28 days with o-anthranilic acid, sulfisoxazole, or sodium chloride had no statistically significant effect on RET^CD59− values (Figs. 9E, 10E, and 11E). RBC^CD59− frequencies were also unaffected by these treatments that included limit dose levels of 1000mg/kg/day (Figs. 9F, 10F, and 11F).

In agreement with Pig-a analyses, RET frequencies associated with MN-RET analyses showed that o-anthranilic acid, sulfisoxazole, and sodium chloride treatments had little-to-no effect on erythropoiesis function. o-Anthranilic acid and sodium chloride showed no evidence of MN-RET induction in 3-day or 28-day studies (Figs. 12I and K). A statistically significant result occurred in the 3-day sulfisoxazole study, that is, %MN-RET was slightly elevated in the high-dose group (2000mg/kg/day; Fig. 12J). However, there are several reasons why we consider this result of little or no biological significance and therefore negative for MN induction—lack of a dose-response relationship; increase less than 2-fold (vehicle-control mean = 0.13% and high dose mean = 0.22%); and the elevated frequency was essentially driven by one rat (0.44%), with the others falling within our range of baseline frequencies.

Relationship Between Pig-a and MN-RET Responses

Each genotoxicant’s maximal mean MN-RET response is graphed against its corresponding maximal mean RBC^CD59− frequency. As shown in Figure 13, there is no obvious relationship between MN induction and Pig-a mutant cell frequency (r ² = 0.005472; p = 0.8391). This lack of agreement is striking considering the fact that the endpoints were measured in the same cell lineage, adding further support to the concept that these assays supply information on two distinct types of DNA damage.

FIG. 13.

Maximal mean MN-RET (%MN-RET) is graphed against its corresponding maximal mean mutant erythrocyte frequency for each of 10 reference genotoxicants that have been studied by this laboratory using the immunomagnetic separation approach to score Pig-a mutant cells. The line and equation correspond to a linear best-fit, and the associated r ² and p values indicate that these endpoints are not highly correlated. Chlor, chlorambucil; Mel, melphalan; Thio, thiotepa; CP, cyclophosphamide; Aza, azathioprine; 2AAF, 2-acetylaminofluorene; HU, hydroxyurea; MMS, methyl methanesulfonate; 1,3-PS, 1,3-propane sultone; BP, benzo(a)pyrene.

DISCUSSION

Data generated to date with erythrocyte-based Pig-a mutation assays are promising. Beneficial characteristics that have become evident include: responsive to reference mutagens (Bhalli et al., 2011b; Cammerer et al., 2011; Lynch et al., 2011; Shi et al., 2011; Stankowski et al., 2011), accumulation of damage with repeat dosing (Bhalli et al., 2011a; Miura et al., 2009), compatibility with commonly used toxicology study designs (Dertinger et al., 2010), high potential as a cross-species–bridging endpoint (Dobrovolsky et al., 2009; Peruzzi et al., 2010; Phonethepswath et al., 2008), and good interlaboratory transferability (Dertinger et al., 2011c). The data presented herein reinforce and extend previous studies and contribute to our knowledge about how the assay behaves when weak mutagens and presumably nonmutagenic agents are tested.

Seven of the eight reference genotoxicants caused significant induction of Pig-a mutant cells. Increased mutant cell frequencies were noted in both 3-day and 28-day study designs. In general, the protracted administration schedule resulted in the greatest effects, and this is likely attributable to the greater total exposure levels that occurred in this design. Induction of mutant cells by the agents studied herein was modest relative to earlier studies with more potent mutagens that included ENU and 7,12-dimethylbenz[a]anthracene. This was expected, as the purpose of these studies was to expand the database of chemicals studied to include weak and nonmutagenic agents.

Past experiences with several of these chemicals makes some of the current results especially noteworthy. In regard to cyclophosphamide, the literature generally shows negative results for mutagenesis in hematopoietic cells of transgenic animals (Gorelick et al., 1999; Lambert et al., 2005). It is possible that negative and weak effects may be explained by differences among reporter genes, because Walker et al. (1999) reported a negative lacI result in mouse lymphocytes, whereas the same animals’ endogenous Hprt locus exhibited a 7.5-fold increase over controls. The Pig-a frequencies induced by cyclophosphamide in the current studies were quite modest. For example, the greatest mutant RBC value observed was 6.3 × 10^–6. Even so, this was a 9-fold increase in mutant RBC frequency relative to the vehicle-control group. We attribute the Pig-a assay’s ability to clearly recognize responses in this range to the immunomagnetic separation procedure, a strategy that provides a reliable estimation of spontaneous mutants through the analysis of many times more cells than is otherwise feasible.

Azathioprine is a hematopoietic carcinogen whose carcinogenic mode of action is uncertain, and that has yielded inconsistent results when studied in in vivo mutation assays. Results to date suggest that azathioprine is negative in rodent Hprt assays (Bendre et al., 2005), although the analysis of Hprt mutants is complicated by the fact that azathioprine is metabolized to 6-thioguanine, a selection agent for mutant cells. Transgenic mouse assays do not suffer from this complication, and the one published study using a transgenic model suggests that the agent is a weak in vivo mutagen (Smith et al., 1999), results that are consistent with the Pig-a data presented herein.

The results with hydroxyurea are also noteworthy, as this was the only presumed genotoxicant that did not affect Pig-a mutant cell frequencies. Hydroxyurea is not DNA reactive, rather its well-established in vitro and in vivo clastogenic activities are thought to be the result of ribonucleotide reductase inhibition (Krakoff et al., 1968). The effects that we observed with the MN-RET endpoint are consistent with that mode of action. The mutagenic potential of hydroxyurea is less clear. In their review of the data, IARC concluded that hydroxyurea does not induce mutations in bacteria or at the Hprt locus of mammalian cells (IARC, 2000). There is a report of hydroxyurea-induced mutation in the pUR288 transgenic mouse model (Martus et al., 1999). The effects reported for lung tissue (6.6 × 10^–5 increased to 18.3 × 10^–5) and lymphocytes (8.9 × 10^–5 increased to 13.7 × 10^–5) were subtle, and statistical analyses were not performed for this experiment that utilized low numbers of rodents (3 in the negative control group and 6 in the hydroxyurea group). Conversely, the Pig-a results reported herein were clearly negative, despite the fact that exposure to the bone marrow compartment was demonstrated by effects on erythropoiesis and by the increased MN-RET frequencies. These new in vivo data may therefore be viewed as reconciling and clarifying previous Ames and other genotoxicity results, adding support to the notion that hydroxyurea’s genotoxic activity is likely to be mediated through disruption of nucleotide pools, causing clastogenic effects at high doses, but with little or no in vivo mutagenic activity, at least in the hematopoietic compartment that is known to be particularly sensitive to this agent.

Each of the eight reference genotoxicants increased the frequency of MN-RET in rat blood. In general, the largest treatment-induced effects were associated with the 3-day studies, where dose/day was higher relative to the dose/day levels that were tolerated in 28-day studies. Even so, it is noteworthy that each of the agents demonstrated significant MN-RET induction in both 3-day and 28-day study designs, adding further support for the ability of the endpoint to be integrated into repeat-dosing schedules.

As depicted in Figure 13, it is interesting that the results of the Pig-a and micronucleus assays were not correlated (in terms of maximal response). As noted above, hydroxyurea in particular showed discrepant results, with significant MN-RET induction but no effect on Pig-a mutation. Another extreme example has been recently described, one that demonstrates the reverse can be true—whereas aristolochic acid showed robust rat erythrocyte Pig-a responses, little-to-no micronucleus induction was evident. This work was performed by FDA-NCTR investigators and was presented at the 2011 Society of Toxicology meeting (Bhalli et al., 2012). A manuscript reporting this finding is in preparation (Bhalli and Heflich, personal communication). Taken together, these data reinforce the complementary nature of the Pig-a and micronucleus endpoints and support the value of integrating gene mutation and cytogenetic damage endpoints into the same study. These data provide important coverage for two distinct modes of genotoxic action that are not realized with any one assay. Importantly, the compatibility of these endpoints with general toxicology studies means that one can collect these genetic toxicology data in the context of studies that greatly aid data interpretation, because pharmacokinetic, histopathology, hematology, and other pivotal data are generated simultaneously. Compatibility with the common laboratory models used in these studies is a distinct advantage of the Pig-a system compared with transgenic mutation models.

Whereas the data presented herein continue to be extremely encouraging, there are several aspects of erythrocyte-based Pig-a assays that require further study. Some of these efforts are currently in progress and include the following: (1) testing of additional chemicals, including mutagenic carcinogens that do not target hematopoietic cells and nonmutagens that exhibit appreciable bone marrow and other toxicities; (2) additional DNA sequencing work to further evaluate the linkage between GPI anchor deficiency and Pig-a gene mutation; and (3) establishment of analogous assays across species of toxicological interest, including the human. With these continuing efforts, erythrocyte-based Pig-a assays should find many roles in product safety assessment programs, regulatory science, occupational exposure situations, and many other circumstances where low-volume blood samples can be made available for study.

SUPPLEMENTARY DATA

Supplementary data are available online at http://toxsci.oxfordjournals.org/.

FUNDING

National Institute of Health/National Institute of Environmental Health Sciences (NIEHS; R44ES018017).