Characterisation of an in vivo Pig-a gene mutation assay for use in regulatory toxicology studies

Abstract

Integration of in vivo genotoxicity testing into standard toxicology studies presents multiple advantages as it reduces animal use and costs, accelerates data generation and provides concurrent data that are useful for interpreting results. The in vivo Pig-a assay is a mammalian gene mutation assay that utilises peripheral blood and thus has a high integration potential. Although inter-laboratory reproducibility has been well demonstrated, further characterisation is required for this assay. In this study, we evaluated intra-laboratory reproducibility of the in vivo Pig-a gene mutation assay (MutaFlow® kit) in rats through the conduct of an assay characterisation prior to subsequent use in Good Laboratory Practices (GLP) toxicology studies. To evaluate intra-laboratory reproducibility, intra-assay and inter-assay variability, ruggedness, robustness and blood storage stability were assessed. These assessments were performed using blood obtained from male Sprague–Dawley rats exposed to 0, 20 or 40mg/kg/day N-ethyl-N-nitrosourea via oral gavage for three consecutive days. Blood was collected from these rats on multiple occasions from Day 29 to Day 71 and samples were analysed for Pig-a mutation using the rat MutaFlow kit. Frequencies of reticulocytes (RET), mutant phenotype erythrocytes (RBC^CD59−) and mutant phenotype RET (RET^CD59−) were determined. Overall, the proportion of RET and frequencies of RBC^CD59− and of RET^CD59− were consistent throughout the different assessments. The assay demonstrated acceptable intra-run and inter-run variability with coefficients of variation of ≤4.8 and 20.6%, respectively. The method was shown to be independent of the analyst performing the assay and unaffected by small changes in assay conditions. Comparable results were obtained from freshly collected samples and samples refrigerated for up to 4 days. Although technically challenging, the rat Pig-a gene mutation assay using a high-throughput automated method was shown to be reliable. The different assay parameters evaluated during the conduct of this study yielded acceptable results. Thus, the method was considered suitable for use in GLP toxicology studies.

Introduction

The integration of in vivo genotoxicity testing into repeat dose toxicology studies in safety assessment offers multiple advantages. It reduces animal use which is in line with the 3Rs principles. It is also cost effective and allows generating data more rapidly compared to expensive transgenic rodent mutation assays. Moreover, it provides concurrent data from multiple endpoints that are useful for interpreting data. One promising assay to detect in vivo gene mutation is the Pig-a assay. This assay that has high integration potential could be used as a follow-up to positive responses in in vitro gene mutation assays, as required by regulatory safety guidelines (1). In addition, this assay that utilises a small amount of blood could easily be combined with other complementary in vivo genetic toxicology endpoints such as the micronucleus and the comet assay (2–6).

The Pig-a assay is based on the detection of cells lacking glycosylphosphatidylinositol (GPI) anchors as a consequence of mutations in the X-linked Pig-a gene. Due to their role in anchoring a variety of proteins to the cell surface, impaired synthesis of GPI anchors can be demonstrated by the absence of surface markers. To discriminate cells with Pig-a mutant or wild-type phenotypes, the assay utilises fluorochrome-conjugated antibodies specific to GPI-anchored markers (7) such as CD59 among other markers. Mutant cells lacking GPI anchors and thus surface CD59 do not bind to conjugated antibodies and thus do not fluoresce. Execution of this assay must be undertaken carefully as sample mislabelling could lead to artificially high mutant frequencies and potentially false-positive results. Sample labelling and processing has in fact been identified by the International Workshops on Genotoxicity Testing Pig-a Workgroup as the potential ‘primary variable influencing the assay outcome across laboratories’ (8). Despite this technical aspect, a high degree of inter-laboratory transferability and reproducibility has been demonstrated in multiple published reports for this assay (2–6,9). However, there is a lack of studies in the literature specifically addressing the intra-laboratory reproducibility (8). Considering our broad experience with flow cytometry methodologies and despite a more narrow experience with the Pig-a assay and in vivo gene mutation assessments, we sought to investigate the reproducibility of this assay.

In the current report, intra-laboratory reproducibility of the assay has been investigated through the conduct of an in-house characterisation study of the rat in vivo Pig-a gene mutation assay. The commercially available rat MutaFlow® kit was used to evaluate the intra-assay and inter-assay variability, assay ruggedness and robustness as well as storage stability of refrigerated whole blood samples.

Negative and positive control samples were generated as means of evaluating these different assay parameters. Rats were acutely exposed to N-ethyl-N-nitrosourea (ENU), a genotoxic compound widely used to induce Pig-a gene mutations in rodents (2,9–12). Peripheral blood samples harvested from these rats were used in the different parameter assessments. The proportion of RET (%RET) as well as frequencies of mutant phenotype RBC (RBC^CD59−) and mutant phenotype RET (RET^CD59−) were compared to predefined acceptance criteria. The collective data are discussed in terms of the rodent erythrocyte-based Pig-a methodology to serve as a reliable assay of gene mutation in the regulatory environment for safety assessment.

Material and methods

Reagents

ENU (CAS no. 759-73-9) was purchased from Sigma–Aldrich, St Louis, MO. Lympholyte®-Mammal cell separation media was purchased from Cedarlane, Burlington, NC. Anti-PE MicroBeads, QuadroMACS™ Separators and LS Columns were purchased from Miltenyi Biotec, Bergisch Gladbach, Germany. Anticoagulant solution, buffered salt solution, nucleic acid dye solution (contains SYTO® 13), anti-CD59-PE and anti-CD61-PE were from In Vivo Rat MutaFlow Kits purchased from Litron Laboratories, Rochester, NY. CountBright™ Absolute Count Beads (Molecular Probes®) were purchased from Life Technologies, Carlsbad, CA. Fetal bovine serum from Hyclone was obtained through Thermo Fisher, Waltham, MA. Cytometer Setup and Tracking and all solutions required for flow cytometer operation were from BD (Franklin Lakes, NJ).

Animals, treatments and sample collections

All animals used in this study were cared for in accordance with the principles outlined in the current ‘Guide to the Care and Use of Experimental Animals’ as published by the Canadian Council on Animal Care and the ‘Guide for the Care and Use of Laboratory Animals’, a National Institutes of Health publication. Protocols were reviewed and approved by the appropriate Institutional Animal Care and Use Committee.

Male Sprague-Dawley rats, ~7 weeks old at receipt, were purchased from Charles River Laboratories (Portage, MI). The rats were housed in groups of up to three in polycarbonate bins in an environmentally controlled room. Temperature and humidity were continuously monitored and recorded. A standard certified commercial rodent chow [Teklad Certified Rodent Diet (W) #8728C; Harlan Laboratories, Indianapolis, IN] and water were available ad libitum throughout the study period, except during designated procedures.

Three groups of 2, 6 and 2 male rats per group were, respectively, administered 0, 20 and 40mg/kg/day ENU. These rats served as negative and positive controls for the testing described below. For ENU treatment, male rats were allowed to acclimate for at least 4 days before the first treatment. The test item was prepared freshly on each day of treatment. ENU was dissolved using Phosphate-Buffered Saline (PBS) pH 6.0 and was administered at 0, 20 and 40mg/kg/day once daily by oral gavage, at ~24-h intervals, for three consecutive days. The dose volume was 10ml/kg for all rats, including controls. The actual volume administered was calculated and adjusted based on the most recent practical body weight.

Approximately 0.3–0.5ml of peripheral blood was collected from each rat by jugular venipuncture using syringes/needles pre-coated with kit-supplied anticoagulant solution. For ENU-treated rats, blood sampling occurred between Day 29 and Day 71 following treatment initiation. The entire volume of blood was immediately transferred into the bottom of an appropriately labelled K₂EDTA tube. Samples that were not analysed within 2h after collection were stored refrigerated.

Sample preparation and data acquisition

For each analysis, 80 μl of blood was transferred into tubes containing 100 μl of kit-supplied anticoagulant solution. Samples were then leukodepleted and processed for Pig-a analysis as per instructions provided with the MutaFlow Kit and as described previously (13,14). Analyses were performed according to the plate-based method in which 96-well microtiter plates were used for the labelling and washing steps with the exception of one robustness assessment in which plate- and tube-based methods were used.

For each blood sample, pre-column and post-column samples were prepared. Pre-column samples consisted of a small volume of cells taken from each labelled sample and stained before going through immunomagnetic separation. Post-column samples consisted of labelled samples stained after immunomagnetic separation. Pre-column samples were acquired for 1min and data were used to calculate %RET as well as RBC to Counting Bead and RET to Counting Bead ratios. Post-column samples were acquired for 3min and data were used to calculate mutant phenotype RBC to Counting Bead and mutant phenotype RET to Counting Bead ratios.

In each individual run, a freshly prepared Internal Calibration Standard (ICS) was included. This ICS consisted of a mixture of leukodepleted, non-antibody-labelled and stained blood sample with an antibody labelled and stained pre-column sample. This mixed sample mimicked the presence of wild-type and mutant cells in both the RET and mature RBC populations. It was used to control the adequacy of experimental settings and of analysis gates.

Data were acquired using a BD FACSCanto II flow cytometer equipped with a carousel loader and controlled by FACSDiva version 6.1.3 software. On each day of acquisition, a calibration of the instrument was performed using the Cytometer Setup and Tracking beads (BD). Flow cytometer experimental set-up was done for each individual run performed. Data analysis was performed with FCS Express version 4 software.

In-house characterisation testing

The intra-assay and inter-assay variability of the method were evaluated using three individual blood samples collected on Day 60: one from a vehicle control and two from ENU-treated (20 and 40mg/kg/day) male rats. These three samples were each split into multiple aliquots. Three aliquots of each sample were processed and analysed in a single run to assess the intra-assay variability. For the inter-assay variability assessment, a total of three independent runs were performed. For each run, different preparations of working reagents and of ICS were used. All three runs were performed in a single day to avoid bias by any potential sample stability issue. The results obtained from the intra-assay variability assessment were considered as the first inter-assay variability run.

The method ruggedness was evaluated by having two analysts each conducting independent runs in which the same set of blood samples collected on Day 60 was processed and analysed. The results obtained by Analyst 1 were considered as benchmark values and were used to compare the values obtained by Analyst 2. For each of the measurements, percent relative error (%RE) was determined.

The robustness of the assay was tested by applying small changes to the experimental conditions to determine whether some factors could impair assay performance. Samples from a vehicle control and an ENU-treated rat (20mg/kg/day) collected on Day 67 were each split into five aliquots and were analysed for Pig-a mutant frequency. One aliquot was analysed using standard experimental conditions to establish benchmark values. The four other aliquots were each analysed with one modification to the procedures. These included an increase in blood sample volume (from 80 to 100 μl), a decrease in antibodies concentration (by 16.7 and 20%, respectively, for anti-CD59 and anti-CD61 antibodies), a modified temperature during incubation with nucleic acid dye solution (from 37°C to room temperature) and an increased incubation time with antibody solution and anti-PE microbeads (from 30 to 60min each). %RET and frequencies of RBC^CD59− and RET^CD59− were calculated and compared to the benchmark values.

Additional robustness testing was conducted by comparing the so-called plate-based method versus the tube-based method. For this assessment, three individual blood samples collected on Day 29 were used: one from a vehicle control and two from ENU-treated rats (20 and 40mg/kg/day). The results obtained with the plate-based method were compared to the values obtained with the tube-based method that were used as benchmark values. %RE was calculated for each measurement.

Sample storage stability was evaluated using blood samples harvested on Day 71 from one vehicle control and two ENU-treated (20 and 40mg/kg) rats. Pig-a analysis was performed on freshly collected blood samples (sample processing started within 2h of collection) and the results obtained served as benchmark (Day 0) values. Blood samples that were immediately stored refrigerated were then analysed at ~24-h intervals for up to 96h. Results obtained on each storage stability time point were compared to the benchmark values.

Calculations and acceptance criteria

The formulas used to calculate RBC^CD59− and RET^CD59− frequencies based on data from pre- and post-immunomagnetic column blood analyses have been described previously (12). Reticulocyte values are expressed as percentages and mutant phenotype RBC and mutant phenotype RET values are expressed as number per 10⁶ total RBCs or total RETs.

Percent coefficient of variation (%CV) and %RE were calculated with the following formulas:

%CV or %RE for RBC^CD59− and RET^CD59− were determined only for samples obtained from ENU-treated rats. %CV and %RE were not calculated for vehicle control samples due to the very low incidence of mutant phenotype cells. In these cases, an absolute number of mutant phenotype cells (≤5×10⁻⁶ mutant cells) was defined as an acceptance criterion based on the literature.

Acceptance criteria for the intra- and inter-assay variability assessments were %CV ≤30% for %RET and ≤25% for RBC^CD59− and RET^CD59−. For the ruggedness, robustness and stability assessments, %RE had to be ≤30% for %RET and ≤25% for RBC^CD59− and RET^CD59−.

Results and discussion

The observed frequencies of RBC^CD59− and RET^CD59− in ENU-treated rats were dependent on the dose administered, as previously reported by others (9). Obtained values were also similar or slightly lower compared to the literature (9,13).

The mean values from the intra- and inter-assay assessments and calculated %CV are presented in Table 1. These results show that intra- and inter-assay %RET measurements resulted in %CV ≤ 4.0 and ≤7.6, respectively. Mean frequencies of RBC^CD59− and RET^CD59− for the vehicle control sample remained ≤0.4 per 10⁶ total RBC and ≤0.6 per 10⁶ total RET in both the intra- and inter-assay variability assessments. Mean frequencies of RBC^CD59− and RET^CD59− for the ENU-treated rat samples resulted in %CV ≤ 4.8 and ≤20.6, respectively, in the intra- and inter-assay variability assessments. The variability obtained in this assessment was within the predefined acceptance criteria.

From the ruggedness results shown in Table 2, the proportion of RET measured in each sample by Analyst 2 was within −13.7 and −9.3 %RE compared to results obtained by Analyst 1. The number of mutant phenotype cells scored in the vehicle control sample remained low for the two analysts with numbers of RBC^CD59− and RET^CD59− ≤ 0.5×10⁻⁶ total RBC and ≤1.8×10⁻⁶ total RET, respectively. In the two ENU-treated rat blood samples evaluated, the frequencies of RBC^CD59− and RET^CD59− measured by Analyst 2 were within 2.4 and 12.2 %RE compared to results obtained by Analyst 1. These results show that different analysts could produce comparable results using the rat MutaFlow® method.

The samples that were analysed using modified experimental conditions showed %RET values within −3.1 and 11.5 %RE compared to the benchmark values established using standard experimental conditions (Table 3). Frequencies of RBC^CD59− and RET^CD59− for the vehicle control sample were ≤0.6 per 10⁶ total RBC and ≤2.0 per 10⁶ total RET, respectively. The frequencies of mutant phenotype RBC were within −16.5 and 4.2 %RE for the ENU-treated rat samples compared to benchmark values. The frequencies of mutant phenotype RET were within −24.1 and 1.5 %RE for the ENU-treated rat samples compared to benchmark values. Additionally, the results were not affected by the use of different assay formats (Table 4). When the so-called tube-based and plate-based methods were compared, %RET values were within −2.2 and 2.5%RE. For the vehicle control sample, frequencies of RBC^CD59− and RET^CD59− were ≤0.2 per 10⁶ total RBC and ≤0.7 per 10⁶ total RET, with both assay formats. Frequencies of RBC^CD59− and RET^CD59− were within −4.4 and 5.1 %RE for both ENU-treated rat samples. Taken together, these results show that the assay had acceptable performance when subjected to different modified experimental conditions and thus the assay was considered robust to these perturbations.

After ~24, 48, 72 and 96h of storage, results obtained for each measurement were consistent with the results obtained from freshly harvested blood. Table 5 shows that in comparison with Day 0 values, %RET were all within −14.3 and 4.2 %RE. The incidence of RBC^CD59− and RET^CD59− obtained from the vehicle control sample remained below 3.2×10⁻⁶ at all tested storage durations. The frequencies measured from the ENU-treated rat blood samples were between −3.3 and 9.5 %RE. Thus, acceptable stability was demonstrated for blood samples stored refrigerated for up to 4 days. These results are consistent with the finding that whole blood samples can be stored refrigerated for several days, as demonstrated by others (8).

Conclusions

Acceptable results were obtained for the parameters tested in this study. We thus showed that this assay provides suitable intra-laboratory reproducibility. We also demonstrated that small changes to the experimental conditions did not affect the assay performance, despite the criticalness of sample labelling and processing. This warrants using well-optimised in vivo Pig-a scoring protocols to generate gene mutation data as part of regulatory toxicology studies.

This in-house method characterisation study was conducted with samples from rats treated with potent doses of a well-established mutagen. Future work should include assessing assay performance at lower levels of mutation.

Acknowledgement

The authors would like to acknowledge the scientific contribution of Stephen D. Dertinger (Litron Laboratories).

Conflict of interest statement: None declared.

References