Tc

LR

a

ARRAA

KMFIB

1

tttAdntasanpt

1h

Mutation Research 752 (2013) 8– 13

Contents lists available at SciVerse ScienceDirect

Mutation Research/Genetic Toxicology andEnvironmental Mutagenesis

j o ur nal homep ag e: www.elsev ier .com/ locate /gentoxCo mm uni t y add re ss : www.elsev ier .com/ locate /mutres

he use of immunomagnetic separation of erythrocytes in the in vivo flowytometer-based micronucleus assay

ilianne Abramsson-Zetterberg ∗, Rickard Carlsson, Salomon Sandisk Benefit Department, The Swedish National Food Agency, Box 622, 751 26 Uppsala, Sweden

r t i c l e i n f o

rticle history:eceived 25 May 2012eceived in revised form 27 October 2012ccepted 15 December 2012vailable online 31 January 2013

eywords:icronucleus

low cytometrymmunomagnetic separationenzo(a)pyrene

a b s t r a c t

The use of sensitive test systems makes it possible to detect weakly genotoxic chemicals and to betterdefine the shape of dose–response relationships, which make it easier to interpret the mechanism behindpossible effects.

In this study we have refined the flow cytometer-based micronucleus assay by use of a cytometerequipped with two lasers. Since micronucleated young polychromatic erythrocytes, MNPCE, are veryfew in number among the cells in peripheral blood, about one or two out of 100,000 erythrocytes, thereis always a risk that other cells, doublets or crystals, by mistake will be classified as a MNPCE. Withimmunomagnetic separation of the very youngest erythrocytes – which are transferrin-positive (Trf + Ret)– prior to analysis, we have obtained an almost pure (>98%) Trf + Ret-population. To clarify whether thisseparation of cells prior to analysis increases the sensitivity of the already sensitive and further refinedflow cytometer-based micronucleus assay, we studied the dose–response towards benzo(a)pyrene, B[a]Pin the low-dose region, 0–30 mg/kg bw. Thirty FVB mice were intraperitoneally injected with B[a]P. Fromthe same blood samples collected from these mice, cells were prepared in the two different ways andanalyzed in the flow-cytometer equipped with two lasers. The lowest dose of B[a]P that can be reliably

determined without being overwhelmed by the estimated error was about the same for the two methods,about 7 mg/kg bw, i.e. the immunomagnetic separation did not increase the sensitivity. A second studywith BalbC mice strengthens the result obtained with the FVB mice. Prior to the low-dose study theoptimal sampling time for the two methods was determined. In this case, the water-solouble chemicalacrylamide was used. The time courses obtained show almost the same shape of the curves, with amaximum of fMNPCE and fMNTrf + Ret at about 40–50 h after exposure.

. Introduction

There are several reasons why it is important to use sensitiveest systems in genotoxicity studies, not least to give the possibilityo detect weakly genotoxic chemicals. More solid data also facili-ate the interpretation of mechanisms behind an obtained effect.lthough clear results are lacking, the general opinion is that theose–effect relationship of genotoxic compounds is linear, i.e. eachew molecule that we are exposed to also increases the risk thathe molecule will bind to DNA, cause a lesion, and end up with

point mutation or a chromosome mutation (single- or double-trand break). As the biomarkers for exposure, e.g., haemoglobinnd DNA adducts, do not always reflect the genotoxic effect, it is

ecessary to obtain information on genotoxicity that is as clear asossible (for a review see Swenberg et al.) [1]. Since there are fac-ors, e.g., DNA repair, that may have an impact on the linearity of

∗ Corresponding author. Tel.: +46 18 17 57 63; fax: +46 18 10 58 48.E-mail address: liab@slv.se (L. Abramsson-Zetterberg).

383-5718/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.mrgentox.2012.12.010

© 2013 Elsevier B.V. All rights reserved.

the genotoxic response, some in vivo studies have been undertakenin order to clarify the shape of the dose–response curve [2–8]. Con-cerning in vivo genotoxicity studies in mammals, it is not easy toclearly demonstrate the shape of the dose–response curve, espe-cially in the low-dose region. There are factors that impact on thestatistical power, e.g., different effects between animals belongingto the same dose group, the number of analyzed cells as well as theproportion of false–positive events, classified as positive.

One of the most common short-term assays for the study ofin vivo genotoxicity is the micronucleus (MN) test. The test withmouse bone-marrow erythrocytes [9,10] is the most developedand most widely used among the MN tests. Since the beginningof the seventies, the method has been further developed. The firstattempt to automate the analysis of micronucleated erythrocyteswas made by Hutter and Stöhr, who used flow-cytometry [11].The analysis was made on a flow cytometer equipped with only

one laser, allowing the discrimination between micronucleatedand non-micronucleated erythrocytes. With the flow cytometerit was, however, not possible to separate older (NCE, normochro-matic) and younger (PCE) erythrocytes. By use of a flow cytometer

L. Abramsson-Zetterberg et al. / Mutation Research 752 (2013) 8– 13 9

F mice,t

e4dstaetTt

dotSaebetttdnIaaTitpwsambsatt

2

2

ccwrtaa

ig. 1. An overview of the two preparations. From the same sample collected fromhe immunomagnetic separation and the Percoll-cleaned-technique.

quipped with two lasers, operating at both multiline UV and88 nm, Grawé et al. [12] demonstrated that it was possible toiscriminate between PCE and NCE, which in turn increased theensitivity. A few years later, in 1996, Dertinger et al. [13] showedhat it was possible to use a flow cytometer with only one laser,nd at the same time distinguish very young, transferrin-positive,rythrocytes (Trf + Ret) from other erythrocytes. The solution waso use antibodies connected to fluorescent dyes specific for therf + Ret. Since then several studies with mice have been publishedhat used both flow-cytometer techniques.

Although the analyses with the two flow-cytometric techniquesescribed above are restricted to the youngest erythrocytes, PCEr Trf + Ret, the cells are, at the time of analysis, mixed with allhe other blood cells, i.e. older erythrocytes and nucleated cells.ince micronucleated young polychromatic erythrocytes, MNPCE,re very few in number, approximately one or two out of 100,000rythrocytes, there is always a risk that subcellular or other debrisy mistake will be classified as micronuleated erythrocytes. How-ver, by simply cleaning the cells through a Percoll gradient prior tohe analysis of MNPCE this problem was reduced [12]. The propor-ion of MNTrf + Ret among all other cells is even lower, roughly lesshan one in a million. To overcome this condition we have furthereveloped the already established flow cytometer-based micro-ucleus assay by using a cytometer equipped with two lasers [12].

n this study, we have mixed whole blood cells from the mice withnti-transferrin antibodies, anti-CD71, bound to magnetic beadsnd thereby captured the cells of interest. All the other cells, notrf + Ret, are then ‘rinsed away’ from the captured cell populationntended for analysis in the flow cytometer. The technique is almosthe same as the one already developed for human erythrocytes,ublished by Abramsson-Zetterberg et al. [14]. We wanted to clarifyhether this newly developed technique further increases the sen-

itivity of the already available flow cytometer-based micronucleusssay. In this study we compare the two techniques. We exposedice to different low concentrations of the genotoxic compound

enzo(a)pyrene (B[a]P), collected blood, and from the same bloodamples prepared the cells in the two different ways. The cells werenalyzed in the flow cytometer equipped with two lasers. Based onhe dose–response curves obtained, we discuss with which statis-ical power one can determine the shape of the curves.

. Materials and methods

.1. Experimental design and sampling

In the already established and several times published MN assay [2–4,12], theells are cleaned through a Percoll gradient prior to the analysis. This Percoll-leaning technique is now compared with the same flow-cytometer technique in

hich the youngest erythrocytes – Trf + Ret – are subject to immunomagnetic sepa-

ation prior to analysis, by use of anti-transferrin antibodies, anti-CD71, connectedo magnetic beads (see Fig. 1). The study is divided in two separate experiments, and b. In both experiments the two techniques are used in parallel, and the resultsre compared.

blood is processed in two different ways prior to the flow-cytometric analysis, i.e.

Experiment (a) was conducted to clarify if the optimal sampling time is the samefor Trf + Ret and PCE. Here we studied the time courses. The mice were given a singlei.p. injection of acrylamide (AA) dissolved in PBS, at a dose of 50 mg/kg bw. Bloodsamples were collected at 0, 20, 30, 35, 45, 50 55, 62, and 75 h after injection. Theblood sampled from each mouse was divided into two parts, one intended for theimmunomagnetic separation method and the other for the Percoll-cleaning tech-nique (Fig. 1). A total of twenty-six male FVB mice, six weeks old, were includedin this experiment. We used three mice per group, except for the sampling timeat 20 h after injection, where two mice were used. Based on the result from thisfirst study a second time course, containing only 12 mice, three per group, exposedto benzo(a)pyrene, B[a]P, was determined. Only the fMNPCE was analyzed in thislimited study where the sampling times were, 0, 47, 63, and 73 h after injection.

Experiment (b) was conducted to clarify which of the two techniques, as illus-trated in Fig. 1, resulted in a dose–response curve with lowest uncertainty. Thirtymale FVB mice, seven weeks old, were divided into 10 groups and treated with dif-ferent low doses of B[a]P (dissolved in corn oil). The dose groups received 0, 3, 6, 9,12, 15, 18, 21, 24, and 30 mg/kg bw. Blood samples were harvested after 47 h. Onemouse in the highest dose group died. In another experiment, male BalbC mice, sixweeks old, were given a dose of B[a]P between 0 and 32 mg/kg bw. In this case onlythe immunomagnetic separation technique was used.

2.2. Animals

Male Balb-C and FVB mice, aged 6–8 weeks and weighing approximately20–25 g, were bought from NOVA SCB-AB, Sollentuna, Sweden. The mice were keptin an animal house at the National Food Agency, Uppsala, Sweden. The mice wereallowed free access to solid food and tap water, and were kept under 12 h of light and12 h darkness. Injections had a distributed volume of 10 �l/gram bw. Temperatureand humidity were adjusted to be optimal for the animals. Blood samples of about200 �l were collected from the orbital plexus in heparinized tubes under anaesthe-sia with Fluothane. Directly after the blood collection, still under anaesthesia, themice were killed by cervical dislocation. The experiment has been reviewed andapproved by the Uppsala Ethical Committee on Animal Experiments, applicationC334/9.

2.3. Chemicals

The origin of the different chemicals were: Acrylamide (CAS 79–06-01) fromMERCK; Benz(a)pyrene from Sigma CAS 50-32-8; Percoll from GE Healthcare Bio-Sciences AB, Uppsala, Sweden. The fluorescent dye Hoechst 33342 from Sigma,Aldrich, Sweden and the dye Thiazole Orange from Molecular Probes, OR, USA. Mag-netic beads, CELLectionTM Biotin Binder Kit, were from Invitrogen Dynal AS, Oslo,Norway; biotin anti-mouse CD71 antibody was from BioLegend, Aachen, Germany;PBS from Statens Veterinärmedicinska Anstalt, Sweden; Glutaraldehyde from TAABLaboratories, England UK; and Foetal Bovine Serum from Lonza BioWhittaker,Verviers, Belgium

2.4. Flow-cytometric analysis without prior immunomagnetic separation

The procedure with the Percoll-cleaning technique has been described severaltimes before [15]. Briefly, five microliter of heparinized whole blood were layered ona 65% Percoll gradient and centrifuged for 20 min, at 600 × g. After centrifugation,the supernatant containing platelets and the majority of the nucleated cells wascarefully aspirated. The remaining cell pellet was suspended in 40 �l PBS and fixedby adding 1.0 ml of a 2% glutaraldehyde solution. The fixed cells were then storedfor a few days at 4 ◦C. On the day of the flow cytometer-analysis the cells werestained with 1.0 ml staining solution per tube. The staining solution contained thefluorescent dyes Hoechst 33,342 (HO) and Thiazole Orange (TO) and was prepared

by adding 500 �l HO-stock (500 �M in distilled water) and 80 �l TO-stock (1 mg/mlmethanol) to 100 ml PBS. To further refine the method, the staining procedure waspartly modified. To some tubes a little more than 1.0 ml of staining solution wasadded, in order to arrive at blood concentrations as similar as possible between thetubes. The colours of the cell solution obtained disclose the relative concentration

10 L. Abramsson-Zetterberg et al. / Mutation Research 752 (2013) 8– 13

Fig. 2. Dot plots of the flow-cytometric analysis of the cells after immunomagnetic separation. These two dot plots illustrate the result of the analysis of samples from onemouse. Each dot in the two figures illustrates the signal from one event. (a) The structure (y-axis) and size (x-axis) of the events are shown. The uppermost population of thea n to tc The RNc n one

op

2

uswtibcmnbpTwptTfTtCacas

2

th

3

pBewtcsRifl

we studied the time course in detail (see Fig. 3). We chose acryl-amide, AA, as a test compound because it is easily soluble andthe genotoxic effect is well documented. After a single i.p. injec-tion with AA at 50 mg/kg bw we studied the appearance of both

Fig. 3. The time courses of the appearance of MNPCE and MNTrf + Ret. The frequency

nalyzed cells, >98%, are marked in the population named P1. The marked populatioells defined in P1 are analyzed on DNA-content (y-axis) and RNA-content (x-axis).

ells contain a micronucleus. The number of remaining NCE is very low, i.e. less tha

f erythrocytes. The staining continued for 45 min at 37 ◦C. From each mouse twoarallel samples were prepared.

.5. Flow-cytometric analysis with prior immunomagnetic separation

The immunomagnetic separation is accurately described in the enclosed prod-ct description of the CELLection Biotin Binder kit. The procedure is also almost theame as we already described for the preparation of human blood [14]. Briefly, 200 �lhole blood was drawn from the mouse and washed twice by mixing PBS and cen-

rifugation. The supernatant was carefully removed and the pellet was re-suspendedn 120 �l PBS containing 2 �l CD-71 mouse antibodies. In the next step the antibody-ound cells were mixed with the magnetic beads. The subsequent selection of theells bound to the paramagnetic beads was achieved by attaching the tubes to aagnetic rack, leaving for five min, and then collecting the liquid – containing the

on-Trf + Ret – by accurate and careful aspiration. To rinse away as much as possi-le the non-Trf + Ret, the tubes were taken from the rack, filled with PBS, and again,laced in the magnetic rack. Again, the non-Trf + Ret was then carefully aspirated.his last step was repeated 3 times. After this immunomagnetic separation, DNAseas added to each tube, in order to release the collected cells from the beads. Therocedure was finished by attaching the tubes again to the magnetic rack to collecthe beads, while the separated and released cells were collected in the PBS solution.he cell solution was centrifuged at 300 × g [please give proper ‘× g’-value here; Ed.]or five min, the supernatant was discarded, and the cell pellet, mainly containingrf + Ret, was fixed in the same way as the cells prepared by the Percoll-cleaningechnique, i.e. one ml of 2% glutaraldehyde fixative was added during vortexing.ells were left in a refrigerator until FACS analysis, and stained just prior to thenalysis in the same way as described for the Percoll-cleaning technique, but in thisase the concentrations of the fluorescent dyes were lower. To 100 ml PBS, 30 �l HOnd 20 �l TO were added, using the same stock solutions as described above. Thetaining continued for 45 min at 37 ◦C.

One parallel sample was prepared from each animal, and analyzed.

.6. Analysis of erythrocytes

The samples were analyzed on a dual-laser FACStar Plus flow cytometer (Beck-on Dickinson, Sunnyvale, CA, USA). The setting and equipment of the flow cytometerave been described several times before, e.g., by Grawé [15].

. Result and discussion

The immunomagnetic separation resulted in isolation of almosture populations of transferring-positive reticulocytes, Trf + Ret.ased on 29 samples prepared from the same number of B[a]P-xposed mice, on average more than 98% of all the separated cellsere defined as Trf + Ret. This calculated proportion is given from

he dot plots showing the size and structure of all the analyzedells, Fig. 2a. In Fig. 2b the cell population defined as P1, Trf + Ret, is

hown, but instead of the size and structure of the cells, the relativeNA- and DNA-content in the defined cells are presented. To ver-

fy the true nature of the presumed MNTrf + Ret, these events wereow-sorted from the population defined as MNTrf + Ret, see Fig. 2b.

he left, P5, represents remaining magnetic beads among the analyzed cells; (b) theA-content discloses the age of the erythrocyte and the DNA-content shows which

per thousand of the total Trf + Ret population.

From five mice a mean of 200 events per mouse were sorted ontoeach of five slides, analyzed in a microscope, and the proportion oftrue events was calculated to be about 98%. Such a pure populationof Trf + Ret is of course an advantage, because the possible contami-nation with, e.g., fragments of other cell types, dublettes, or crystalsimproperly defined as MNTrf + Ret, decreases.

With the technique described above, i.e. immunomagnetic sep-aration prior to the flow-cytometric analysis, the sensitivity ofthe micronucleus assay was evaluated and compared with thetechnique where the blood is prepared with the Percoll-cleaningtechnique. Whatever method is used, one of the prerequisites forhigh sensitivity is an optimal sampling time after exposure. Sincethe CD71 antigen is expressed on the erythrocyte membrane onlyfor a few hours after the cell enters the peripheral blood sys-tem, it seems likely that the appearance of induced MNPCE andMNTrf + Ret will be different, based on the proportion of the amountof obtained Trf + Ret and the amount of whole blood. This is why

of micronucleated polychromatic erythrocytes, fMNPCE (�) and micronucleatedtransferring-positive erythrocytes, fMNTrf + Ret (×) in peripheral blood as a func-tion of time after a single acute intraperitoneal injection of 50 mg/kg bw acrylamide.Animals from an inbred strain of mice were used, male BalbC. Each sampling timerepresents three mice, except at 20 h after injection, where only two mice were used.

L. Abramsson-Zetterberg et al. / Mutation Research 752 (2013) 8– 13 11

Fig. 4. Dose response of benzo(a)pyrene in comparison with other results. The fre-quency of micronucleated polychromatic erythrocytes in peripheral blood, fMNPCE(‰), from FVB male mice exposed to different low doses of benzo(a)pyrene (�).Each dose represents results from three mice, except for the highest dose wheretwo mice were used. From each mouse a mean of 200,000 PCE were analyzed. Ints

M(PAnsadMopctipa

istfidlto76twao[paw

sFrhodliob

Fig. 5. A detailed analysis of the dose–response data obtained. The dose–responserelationship is shown of FVB male mice injected (i.p.) with different low doses ofbenzo(a)pyrene, viz. 0, 3, 6, 9, 12, 15, 18, 21, 24, and 30 mg/kg bw. Linear modelswere fitted to data on fMNPCE (a) and fMNTrf + Ret (b). The two-sided 90% con-fidence intervals on absolute response (fMN) around the fitted linear models areillustrated by the two pairs of curved lines. This confidence interval is estimated bythe profile-likelihood method. The large circles represent observed mean responses(in this example, for 10 different doses, where n = 3 for each dose, except for thehighest dose where n = 2) with 90% confidence intervals. In Sand et al., [24] theconcept of the signal-to-nose crossover dose (SNCD) was introduced. The SNCD wasregarded as the lowest reliably estimated dose without having the signal (estimatedresponse) overwhelmed by noise (estimated error). The SNCD is illustrated for thepresent data, and defined as the dose (6.3 and 7.1 mg/kg bw, respectively) at whichthe additional response (e.g., S = 3.8–2.7 = 1.1 units for fMNTrf + Ret) is equal to theabsolute difference between the upper and lower bound on absolute response (e.g.,N = 4.3–3.2 = 1.1 units for fMNTrf + Ret). For both datasets, the SNCD corresponds toa signal-to-noise ratio (S/N) equal to 1. Apparently, the SNCD becomes similar forthe two datasets, indicating that the ability to detect a response in the low-dose

dose–response relationship is to compare the obtained increase

he same figure, the results from some other studies, published by others, are alsohown, × refers to [21], + corresponds to [22], and − to [23].

NTrf + Ret and MNPCE at the same time from the same miceFig. 3). The result shows that the maximum number of the fMN-CE and fMNTrf + Ret is about the same at 40–50 h after injection.ccordingly, the same sampling time can be used in the two tech-iques (Fig. 3). This information is important when comparing theensitivity of the two techniques. Several studies in this area havelready been published. However, none of these has compared inetail the time schedule of the two cell populations, MNPCE andNTrf + Ret, at the same time from the same animals. The rec-

mmended sampling time for the occurrence of MNTrf + Ret ineripheral blood is generally 48 h after exposure [16–18], and theorresponding recommended sampling time for MNPCE is abouthe same [19,20]. Since the life span of Trf + Ret in peripheral bloods much shorter than the occurrence of PCE it was somewhat sur-rising that the slopes of the two curves of the occurrence of MNPCEnd MNTrf + Ret were almost the same (Fig. 3).

The overall goal for this investigation was to clarify ifmmunoseparation of the youngest erythrocytes prior to analy-is also resulted in higher sensitivity. For this purpose we choseo study the genotoxic effect of B[a]P in the low-dose region. Tond out whether the sampling time to observe a maximum of fMNiffers from the obtained result of AA-exposed mice, we made a

imited study of the occurrence of MNPCE after a single i.p. injec-ion of B[a]P. We studied twelve B[a]P-exposed mice given a dosef 30 mg/kg bw at four different sampling times, viz. 0, 47, 63, and3 h after injection. The corresponding mean fMNPCE were 3.5,.2, 4.6, and 4.1‰. This result indicated that also after i.p. injec-ion of B[a]P the optimal sampling time is not delayed comparedith mice exposed to AA. Earlier published results have shown that

t high doses of an agent, the optimal sampling time for the studyf micronucleated PCE, is delayed in comparison with lower doses19]. This in turn points to the difficulties, in a single study, to com-are the genotoxic effect after exposure to different dose levels over

wide dose range. This was one of the reasons why in this studye restricted the analysis to a narrow dose-range.

In concordance with the result described above, we used a 47 hampling time in the subsequent dose–response study with B[a]P.rom each mouse about 200 �l whole blood was drawn, whichesulted in a mean number of 1–2 million Trf + Ret. In this study,owever, we restricted the number of analyzed Trf + Ret to a meanf 220,000 per mouse and a mean of 200,000 PCE per mouse. Theose interval was 0–30 mg/kg bw B[a]P, which is low in corre-

ation to other published results. The difference in dose interval

n our study compared with what has already been published bythers is demonstrated in Fig. 4. In Fig. 5a and b the results ofoth the blood preparations, the Percoll-cleaning technique and the

region is similar. More generally, the SNCD concept illustrates that for any datasetstatements regarding the response or shape of the dose–response curve below acertain dose level are not meaningful.

immunomagnetic isolation, are presented in detail. The uncertaintyin the results can be expressed in different ways. A simple wayis to calculate variations or standard deviations, S.D., at differentdose levels. When comparing the two techniques of blood prepa-ration, no significant difference was found in the mean S.D. at thedifferent doses. Another way of describing the uncertainty of a

of the fMN response to the uncertainty expressed as a confidenceinterval (estimated error), i.e. the lowest dose that can be reli-ably estimated without the mean effect being overwhelmed by the

12 L. Abramsson-Zetterberg et al. / Mutat

FaT

elt(se

ff0Trt

milabrgtrccdci

ioitsiimitwteie

4

io

[

[

[

[

[

[

[

[

ig. 6. The dose–response relationship in BalbC male mice. The study was done inlmost the same way as presented in Fig. 5, but only the fMNTrf + Ret was calculated.he different doses of benzo[a]pyrene were 0, 4, 8, 10, 12, 16, 24, and 32 mg/kg bw.

stimated error (see Fig. 5) [24]. In this study we found that theowest dose, about 7 mg/kg bw, was approximately the same forhe two curves, the two techniques have about the same sensitivityFig. 5). The high number of analyzed cells and the refinement of thetaining of the cells in the Percoll-cleaning technique are reliablexplanations for the similar sensitivities of the two techniques.

To find out if there were differences in the sensitivity of dif-erent mouse strains we repeated the dose response study of theMNTrf + Ret in male BalbC mice after i.p. injections of B[a]P at–32 mg/kg bw (Fig. 6). About the same number of mice was used.he resulting dose response was almost the same, and the lowesteliable estimated dose was also about the same as calculated forhe FVB mice.

The shape of the dose–response curves can be described withany different mathematical equations, of which the simplest one

s the linear model, y = l + kx. In this study we assumed that theinear model was appropriate for the fMN data that we obtained,lthough the power model, y = l + kxn, fitted the data significantlyetter (data not shown). However, this result was due to a singleesponse value in the highest dose group, which deviated from theeneral trend (Figs. 4 and 5). Moreover, taking into considerationhe data obtained from the BalbC mice, a linear relationship is moreeliable (Fig. 6). During the last years several publications have dis-ussed the shape of the dose–response relationship for genotoxicompounds. The studies showing no increased effect in the lowestose region of the well-known compound ethylmethane sulfonateontributed strongly to this discussion (see a special issue concern-ng this topic [25]).

Irrespective of which technique is used – sensitive or not –, theres always a lowest limit of detection where the confidence intervalverwhelms the calculated increased dose response. In spite of this,t is of course an advantage to reach as high a sensitivity as possible,o minimize uncertain interpretations of dose–response relation-hips. In this study, which used documented sensitive methods,t is clear that the response differences between mice receiv-ng the same doses are a limiting factor. When the number of

ice increases, the accuracy in the calculated dose response alsoncreases. In an earlier published study with the same MN assay,he dose–response relationship at doses as low as 0–18 mGy/dayas calculated. Five mice per dose level were used and thereby

he results clearly indicated a linear relationship [2]. One of sev-ral possible explanations to the different response between mices likely an uneven body distribution of a water-insoluble chemical,.g., B[a]P.

. Conclusion

In this study we first demonstrated that as a result of anmmunomagnetic separation technique to isolate Trf + Ret, we havebtained an almost pure suspension of the cells of interest for

[

ion Research 752 (2013) 8– 13

analysis of the MNTrf + Ret frequency. We analyzed several hun-dred thousand of Trf + Ret per mouse, but the procedure did notsignificantly increase the sensitivity compared with the alreadydeveloped and published method, i.e. the flow cytometer-basedmicronucleus assay with a cytometer equipped with two lasers.Instead, such a pure cell population of Trf + Ret spread out onslides ought to increase the sensitivity of the micronucleus assayperformed in the most common way, i.e. by evaluation via themicroscope. But, on the other hand, the benefits of automation,objective and extremely rapid, are not really addressed by enrich-ment followed by microscopy.

Conflict of interest

None.

Acknowledgements

We thank Elvy Netzel for technical assistance and Prof. em. GöstaZetterberg for valuable discussions. We thank the Swedish CivilContingencies Agency (MSB) for financial support.

References

[1] J.A. Swenberg, E. Fryar-Tita, Y.C. Jeong, G. Boysen, V.E. T- Starr, R.J. Walker,Albertini, Biomarkers in toxicology and risk assessment: informing criticaldose–response relationships, Chem. Res. Toxicol. 21 (2008) 253–265.

[2] L. Abramsson-Zetterberg, J. Grawé, G. Zetterberg, Flow cytometric analysis ofmicronucleus induction in mice by internal exposure to 137 Cs at very low doserates, Int. J. Radiat. Biol. 67 (1995) 29–36.

[3] J. Grawé, L. Abramsson-Zetterberg, G. Zetterberg, Low dose effects of chemicalsas assessed by the flow cytometric in vivo micronucleus assay, Mutat. Res.405–412 (1998) 199–208.

[4] L. Abramsson-Zetterberg, The dose–response relationship at very low doses ofacrylamide is linear in the flow cytometer-based mouse micronucleus assay,Mutat. Res. 535 (2003) 215–222.

[5] M. Hoshi, K. Morimura, H. Wanibuchi, M. Wei, E. Okochi, T. Ushijima, K. Takaoka,S. Fukushima, No-observed effect levels for carcinogenicity and for in vivomutagenicity of a genotoxic carcinogen, Toxicol. Sci. 81 (2004) 273–279.

[6] N. Asano, D.K. Torous, C.R. Tometsko, S.D. Dertinger, T. Morita, M. Hayashi,Practical threshold for micronucleated reticulocyte induction observed for lowdoses of mitomycin C, Ara-C and colchicines, Mutagenesis 21 (2006) 15–20.

[7] E. Zeiger, L. Recio, T.R. Fennell, J.K. Haseman, R.W. Snyder, M. Friedman, Inves-tigation of the low-dose response in the in vivo induction of micronuclei andadducts by acrylamide, Toxicol. Sci. 107 (2009) 247–257.

[8] E. Gocke, L. Müller, In vivo studies in the mouse to define a threshold for thegenotoxicity of EMS and ENU, Mutat. Res. 678 (2009) 101–107.

[9] W. Schmid, Chemical mutagen testing on in vivo somatic mammalian cells,Agents Actions 3 (1973) 77–85.

10] J.A. Heddle, A rapid in vivo test for chromosomal damage, Mutat. Res. 18 (1973)187–190.

11] K.-J. Hutter, M. Stöhr, Rapid detection of mutagen-induced micronucleatederythrocytes by flow cytometry, Histochemistry 75 (1982) 353–362.

12] J. Grawé, G. Zetterberg, H. Amnéus, Flow-cytometric enumeration of micronu-cleated polychromatic erythrocytes in mouse peripheral blood, Cytometry 13(1992) 750–758.

13] S.D. Dertinger, D.K. Torous, K.R. Tometsko, Simple and reliable enumeration ofmicronucleated reticulocytes with a single-laser flow cytometer, Mutat. Res.371 (1996) 283–292.

14] L. Abramsson-Zetterberg, G. Zetterberg, M. Bergqvist, J. Grawé, J. Humancytogenetic biomonitoring using flow-cytometric analysis of micronuclei intransferrin-positive peripheral blood reticulocytes, Environ. Mol. Mutagen. 36(1999) 22–31.

15] J. Grawé, Flow cytometric analysis of micronuclei in erythrocytes, Methods Mol.Biol. 291 (2005) 69–83.

16] G.M. Hynes, D.K. Torous, C.R. Tometsko, B. Burlinson, D.G. Gatehouse, The singlelaser flow cytometric micronucleus test: a time course study using colchicineand urethane in rat and mouse peripheral blood and acetaldehyde in rat periph-eral blood, Mutagenesis 17 (2002) 15–23.

17] M. De Boeck, B.J. van der Leede, F. Van Goethem, A. De Smedt, M. Steemans,A. Lampo, P. Vanparys, Flow cytometric analysis of micronucleated reticulo-cytes: Time- and dose-dependent response of known mutagens in mice, using

multiple blood sampling, Environ. Mol. Mutagen. 46 (2005) 30–42.

18] S.D. Dertinger, Y. Tsai, I. Nowak, O. Hyrien, H. Sun, J.C. Bemis, D.K. Torous, P.Keng, J. Palis, Y. Chen, Reticulocyte and micronucleated reticulocyte responsesto gamma irradiation: dose–response and time–course profiles measured byflow cytometry, Mutat. Res. 634 (2007) 119–125.

Muta

[

[

[

[

[

[

L. Abramsson-Zetterberg et al. /

19] L. Abramsson-Zetterberg, G. Zetterberg, J. Grawé, J. The time course of micronu-cleated polychromatic erythrocytes in mouse bone marrow and peripheralblood, Mutat. Res. 350 (1996) 349–358.

20] OECD Guidelines for the Testing of Chemicals/Section 4: Health Effects TestNo. 474: Mammalian Erythrocyte Micronucleus Test, OECD in Library, (1997).ISBN: 9789264071285.

21] U. Kliesch, I. Roupova, I.D. Adler, Induction of chromosome damage in mousebone marrow by benzo(a)pyrene, Mutat. Res. 102 (1982) 265–273.

22] T. Awogi, T. Sato, Micronucleus test with benzo[a]pyrene using a single peroraladministration and intraperitoneal injection in males of the MS/Ae and CD-1mouse strains, Mutat. Res. 223 (1989) 353–356.

[

tion Research 752 (2013) 8– 13 13

23] C.L. Lemieux, G.R. Douglas, J. Gingerich, S. Phonethepswath, D.K. Torous, S.D.Dertinger, D.H. Phillips, V.M. Arlt, P.A. White, Simultaneous measurementof benzo[a]pyrene-induced pig-a and lacZ mutations, micronuclei and DNAadducts in MutaTM mouse, Environ. Mol. Mutagen. 52 (2011) 756–765.

24] S. Sand, C.J. Portier, D. Krewski, A signal-to-noise crossover dose as the pointof departure for health risk assessment, Environ. Health Perspect. 119 (2011)

1766–1774.

25] W. Lutz (Ed.), Assessment of human toxicological risk of viracept patientsaccidentally exposed to ethyl methanesulfonate (EMS) based on preclinicalinvestigations with EMS and ethylnitrosourea, Tox. Lett. 190 (2009) 239–340(special issue).