The Pig-a Gene Mutation Assay in Mice and Human Cells: A Review

(1)

MiniReview

The Pig-a Gene Mutation Assay in Mice and Human Cells:

A Review

Ann-Karin Olsen^1,2, Stephen D. Dertinger³, Christopher T. Kruger€ ^4,†, Dag M. Eide^2,5, Christine Instanes^1,2, Gunnar Brunborg^1,2,‡, Andrea Hartwig⁴and Anne Graupner^1,2

1Department of Molecular Biology, The Norwegian Institute of Public Health, Oslo, Norway,²Centre for Environmental Radioactivity (CERAD CoE), Norway,³Litron Laboratories, Rochester, NY, USA,⁴Food Chemistry and Toxicology, Karlsruhe Institute of Technology (KIT), Karlsruhe,

Germany and⁵Department of Toxicology and Risk, The Norwegian Institute of Public Health, Oslo, Norway (Received 19 December 2016; Accepted 1 May 2017)

Abstract: This MiniReview describes the principle of mutation assays based on the endogenous Pig-a gene and summarizes results for two species of toxicological interest, mice and human beings. The work summarized here largely avoids rat-based studies, as are summarized elsewhere. ThePig-agene mutation assay has emerged as a valuable tool for quantifyingin vivoand in vitromutational events. ThePig-alocus is located at the X-chromosome, giving the advantage that one inactivated allele can give rise to a mutated phenotype, detectable by multicolour flow cytometry. Forin vivostudies, only minute blood volumes are required, making it easily incorporated into ongoing studies or experiments with limited biological materials. Low blood volumes also allow individuals to serve as their own controls, providing temporal information of the mutagenic process, and/or outcome of intervention. These characteristics make it a promising exposure marker. To date, the Pig-agene mutation assay has been most commonly performed in rats, while reports regarding its usefulness in other species are accumulating. Besides its applicabil- ity toin vivostudies, it holds promise for genotoxicity testing using cultured cells, as shown in recent studies. In addition to safety assessment roles, it is becoming a valuable tool in basic research to identify mutagenic effects of different interventions or to understand implications of various gene defects by investigating modified mouse models or cell systems. Human blood-based assays are also being developed that may be able to identify genotoxic environmental exposures, treatment- and lifestyle-related factors or endogenous host factors that contribute to mutagenesis.

Paroxysmal nocturnal haemoglobinuria (PNH) is a genetic dis- order that affects 1–10 per million individuals and is caused by somaticPIG-Agene mutations within a bone marrow stem cell [1]. ThePIG-Agene encodes a catalytic subunit of the N- acetylglucosamine transferase complex involved in the synthesis of glycosylphosphatidylinositol (GPI) cell surface anchors [2]. Inactivating mutations inhibit the biosynthesis of GPI anchors, causing a deficiency in GPI-anchored proteins on the cell surface, for example CD59, CD55 and CD24. PNH usually affects erythrocytes, granulocytes and monocytes. In a minority of cases, the lymphocyte lineage is also affected. As a cell surface phenotype present on readily accessible haematopoietic cells, flow cytometry-based techniques that measure the frequency of CD59- and/or CD55-deficient red blood cells have replaced the traditional HAM test (acid- induced disruption of the fragile red blood cells of patients with PNH) for PNH diagnosis [3,4]. Work directed at elucidat- ing the aetiology of PNH and creating flow cytometry-based diagnostic tests provided an important starting point for the

development of mutation assays that utilize the Pig-a/PIG-A loci as reporters of gene mutation.

The earliest reports that suggested that PIG-A may repre- sent a useful reporter of gene mutation were from the group of Luzzatto [5] and later Chen [6]. These groups focused on human cells, often transformed lymphocytes grown in cell culture, or freshly isolated blood leucocyte population(s) [5–

9]. Several extended this work to rodent blood erythrocytes and/or reticulocytes in order to provide regulatory safety assessment laboratories with a practical and efficient platform for evaluating chemical or physical agents’ potential to cause gene mutationin vivo[10,11]. To date, this work has primar- ily been accomplished in rats, as a key rodent model used in regulatory toxicology studies [12,13]. However, the success- ful application of phenotypic Pig-a-mutant scoring methods in cell lines and mouse models is important given their prominent role in both regulatory toxicology and basic research. As much understanding has been accomplished and described for rats, the purpose of this MiniReview was to first introduce the Pig-a assay in detail followed by a description of the current status for cell lines, mouse models and human beings.

The scientific literature was searched in PubMed using key- words such as Pig-a, PIG-A, mutation, paroxysmal nocturnal haemoglobinuria, mouse, human beings, TK6 and followed by Author for correspondence: Ann-Karin Olsen, Department of Molecu-

lar Biology, The Norwegian Institute of Public Health, PO Box 4404, N-0403 Oslo, Norway (e-mail ann.karin.olsen@fhi.no).

†Present address: Department of product safety, Beiersdorf AG, Unnastra. 48, 20245 Hamburg, Germany.

‡Retired.

(2)

manual selection of relevant articles using a version of the Pig-a/PIG-Aassays as main inclusion criteria. The studies are presented in tables 1 and 2 according to the publication year.

Principle of the Assay

Iida et al. [14] isolated the human PIG-A gene and found that it contains six exons over its 17 kb length. As demonstrated by Kawagoe et al. [15], there is a high degree of interspecies conservation of the gene’s structure, function and position on the X-chromosome. The gene product plays a critical role in the first step in GPI anchor biosynthesis, and the entire process is thought to require at least two dozen genes. Mutation of any one of these genes could theoretically result in GPI anchor deficiency. However, all other genes involved in GPI anchor synthesis are autosomal. Mutations on both alleles would have to occur to ablate expression of GPI anchors, and this is expected to be an exceptionally rare event, at least when all other GPI anchor-associated gene alleles are both functional. Thus, the single-copy status of the Pig-a/PIG-Agene is a critical feature that makes it useful as a phenotypic reporter of gene mutation, because only one mutational event is sufficient to completely ablate expression of GPI anchors.

The evidence that GPI anchor deficiency represents a valid reporter ofPig-a/PIG-A mutation starts with investigations of PNH. One key finding is that with only extremely rare excep- tions PNH clones exhibit mutation at the PIG-A locus. For instance, Nishimuraet al.[16] analysed 146 patients with PNH and reported the types of mutations that lead to GPI anchor deficiency, all within the PIG-A locus. Single-base substitu- tions and frame-shift events were the most highly represented classes of mutations observed. They also found three examples of large deletions (entire gene, 4 kb and 737 base pairs), as well as a large insertion (88 base pairs). The mutations were widely distributed in the coding regions and splice sites, although Nafaet al. [17] found a somewhat higher frequency of missense mutations in exon 2 relative to other exons. Taken together, the PNH literature provides strong support for the concept that the GPI anchor-deficient phenotype represents a useful reporter of Pig-amutations, and such assays would be sensitive to several important classes of mutations.

The cell surface of haematopoietic cells is dramatically affected by Pig-a/PIG-A mutations. For example, CD59, CD55 and/or CD24 are highly expressed on haematopoietic cells, and these expression levels are maintained throughout cells’ lifespan. However, when bone marrow cells experience an inactivatingPig-a/PIG-Amutation, after a sufficient period of manifestation time, progeny cells show a complete absence of these markers. The significant changes to specific cell surface markers as a result of Pig-a/PIG-Amutations are readily detected using fluorescent antibodies in conjunction with flow cytometry. Furthermore, blood cells as well as lymphoblastoid cells in culture are highly compatible with flow cytometric analysis. These characteristics combine to make flow cytometry the method of choice for efficiently scoring Pig-a/PIG-A- mutant cells (fig. 1).

Variations in the basic approach of scoring mutant cells (mostly blood cells) via flow cytometry in conjunction with fluorescent antibody labelling have been described. For instance, for some cell types, it is possible to use the fluorescent-labelled aerolysin reagent (FLAER) instead of antibody(ies). Whereas antibodies recognize wild-type cells via their recognition of GPI-anchored protein(s), FLAER binds with high affinity to GPI anchors themselves. When the frequency of spontaneous mutant phenotype cells is very low, as in the case for rat and mouse erythrocytes and reticulocytes, the use of immunomagnetic separation technologies is convenient. In this approach, the number of reticulocytes evaluated for the mutant phenotype is enriched, thus decreasing the data acquisition time [18].

Alternatively, one may use immunomagnetic separation to specifically deplete the wild-type cells in each sample. In this scenario, flow cytometric analysis of immunomagnetically sep- arated samples provides the means to efficiently evaluate orders of magnitude more erythrocytes and reticulocytes for the mutant phenotype than would otherwise be possible, leading to lower scoring errors [19]. These immunomagnetic separation techniques can be especially important when study designs seek to keep treatment group sizes modest while at the same time ensuring sufficient statistical power to detect modest changes to spontaneous mutant cell frequencies. This experi- mental refinement greatly adds to the sensitivity of the assay, due to the enrichment of unstained mutants and thus analysis of manyfold more potential mutant cells (fig. 2).

One important aspect of the assay is that the mutants scored by antibody staining and flow cytometry are phenotypical mutants. Verification of these phenotypic mutants as truePig- a/PIG-A mutants by DNA sequencing analyses is vital for confirmation of mutation frequencies measured and to demon- strate the true nature of the mutants. Such identification has been demonstrated in bone marrow erythroids of mice [20]

and CD48-negative T cells of N-ethyl-N-nitrosourea (ENU)- and 7,12-dimethylbenz[a]anthracene (DMBA)-treated rats [21,22]. The latter was performed by flow cytometric sorting of mutant cells and sequencing of thePig-agene of expanded single cells [21,22]. The induced mutation spectra obtained were consistent with expected spectra obtained in other endogenously expressed genes, verifying the validity of the mutation frequencies measured. In human cells, point mutations as well as deletions in PIG-A have been identified in TK6 cell lines [23,24]. Recently, heterogeneous pools ofPig- a-mutant T cells derived from DMBA-exposed rats were efficiently sequenced by an elegant novel technique named mutation analysis with random DNA identifiers (MARDI), requiring no clone-by-clone analyses [25]. The MARDI technique facilitates verification ofPig-amutants in a more feasi- ble manner than previous strategies. Using the MARDI technique, nearly all previously found Pig-a mutations were identified, and new mutations were detected.

Pig-ain Mice

Searching the scientific literature, we identified 21 studies in mice (summarized in table 1). In 2008, the first study to

PIG-AIN MICE AND HUMAN CELLS 79

(3)

Table1. OverviewofPig-astudiesinmice. Strain(s),genotype(s), sex,ageAgent/intervention(exposure,route,assessmenttime).Methodological detailsMainoutcome,mutationfrequencies(Mf)inreticulocytes(RETs)orredbloodcells (RBCs)References CD-1females, 7–8weeksDMBA,ENU(threedailydoses;i.p.;DMBA:75mg/kg/day, analyses3weeksafterexposure;ENU:40mg/kg/day,analysed weekly) IncreasedMfsinRETsandRBCs.Resultsfromweek2.ENU:RETmeanMf 6009106(~43-fold)andRBCmeanMf3009106(~21-fold),maximumeffectin RETsinweek1andRBCinweek2;DMBA:RETmeanMf2869106(~21-fold) andRBCmeanMf1339106(~10-fold),vehiclecontrolmeanMf14219106

[26] C57BL/6males, 6–7weeksENU(i.p.;singledose:0,10,25,45,70,100and140mg/kg;single versussplitdoses:fourweeklydosesversusonesingledose: 498mg/kgversus32mg/kgand4940mg/kgversus 160mg/kg;assessmentwith2-to6-weekintervalsfor 26weeksafterlastexposure)

Singledose:increasedRETandRBCmeanMfswithmaximumeffectatweek2 (RETs,7009106)and4(RBCs,2309106)weeksthatwasmaintainedformany weeks,anddeclinedatweeks20and26.VehiclecontrolmeanMfsrangedfrom0.2 to2.29106(RETs)and0–29106(RBCs).Splitversussingledoses:groups receivingsingledoseshadgenerallyhighermeanMfsthanequaltotalsplitdoses. IncreasedRETmeanMfsat2weeks(32versus894mg/kg:180versus1059106; 160versus4094mg/kg:1150versus7509106 )andRBCmeanMfsat4weeks (32versus894mg/kg:66versus339106 ;160versus4094mg/kg:460versus 4009106 ).VehiclecontrolmeanMfsrangedfrom0to5.59106 (RETs)and 0.6–2.79106 (RBCs) [27] CD-11 males, 6weeksENU(singledoseof100mg/kg,i.p.).Mfmeasuredinerythroids frombonemarrow(BM)andperipheralblood(PB)analysedpre- dosingand1,2and4weeksafterdosing.)

IncreasederythroidMfinBMoccurredearlierandwashigherthaninPB.TheMfsin PBandBMerythroidsreached1999106and6829106at4weeksafter exposure,respectively.Mfincontrolmicewaslow(usually<59106inPBand 0–83.39106inBM)

[20] MutaTMMouse males,25weeksBaP(positivecontrolENU)(dosedwith0,25,50and75mg/kg/day for28days,oralgavage,analysed3daysafterthelastdose; positivecontrolENUwasgivenasasingledoseof45mg/kgbw. i.p.,andanalysed2weeksafterdosing)

Significantdose-dependentincreasesinRETandRBCMfs.AtthehighestBaPdose (75mg/kgbw/day),theRETandRBCmeanMfswere~2409106and ~959106.VehiclecontrolmeanMfswere0inRETsand0.49106inRBCs.Mf inENU-positivecontrolmicewas~1909106(RETs)and~309106(RBCs)

[33] C57BL/6Jmales, 6weeksX-rays(160kVpwith0.5-mmCuand0.5-mmAlfilters)(0.52 Gy/min;singledoseof0.5,1and2Gyorfractionateddoseof 490.5Gyonceaweek;singleexposures:analysesondays0,2, 7,14,21,28,35,42,49,56,70,84,105,126,148,203,232,267, and297afterexposure,forfractionatedexposureanalysesitwas performedjustbeforeeachexposureondays0,7,14and21)

Dose-dependentincreaseinRBCMfandincreaseinRETMf,whichovertimereturned tobackgroundlevels.Equivalentsingleandfractionateddosesgaverisetosimilar maximumRBCmeanMfs;however,themaximumofthefractionateddoseappeared twotothreeweekslaterthanthesingledose.Atthehighestsingledoseof2Gy,the maximumRETandRBCmeanMfswereonday14with44.899106and ~1509106 .Aftertheprotracteddoseof490.5Gy,themaximumRBCmeanMf wasatday42with49.979106 ,whereasthemeanMfofRETsvariedconsiderably probablyduetobonemarrowtoxicity.Thereweremarkedinterindividualdifferences inresponse,bothinthemagnitudeofthemaximalresponseandintimeuntil maximumresponsewasevident

[35] B6C3F1males, 6–7weeksTiO2-NP(positivecontrolENU) (i.v.;exposurefor3dayswith0.5,5.0and50mg/kg;positive controlENUwasgivenasasingledoseof140mg/kgbw.i.p.; analysesoccurredonday1andweeks1,2,4and6afterdosing)

NoincreasedMfinRETsorRBCs,despiteavailabilitytothebonemarrowandbone marrowcytotoxicity.Overthestudyperiod,RETandRBCmeanMfsinvehicle controlanimalsrangedfrom0to1.29106and0–49106.Nosignificantincrease wasobservedintreatedanimals.PositivecontrolanimalstreatedwithENU (140mg/kg)hadmeanRETandRBCMfsat2weeksof864.009106and 304.809106,respectively

[38] (continued)

80 ANN-KARIN OLSENET AL.

(4)

Table1.(continued) Strain(s),genotype(s), sex,ageAgent/intervention(exposure,route,assessmenttime).Methodological detailsMainoutcome,mutationfrequencies(Mf)inreticulocytes(RETs)orredbloodcells (RBCs)References C57BL/6Jgptdelta males,8weeksENU,BaP,4NQO(singleoraldosesofENU(40mg/kg),BaP(100 and200mg/kg)and4NQO(50mg/kg);analysesoccurredat2,4 and7weeksafterthetreatment) IncreasedRBCMfforallthreemutagens,withENUgivinganincreasethatwasstable untilweek7,BaPshowedadeclineafter2weeksand4NQOgivingrisetovariable, lowresponse.At2weeks,theRBCmeanMfswere23.006.969106forENU, 9.252.639106for100mg/kgBaP,15.808.209106for200mg/kgBaP and2.604.729106for4NQO.PBSvehiclecontrolmeanMfswere0.4– 0.89106

[30] CD-1males,7weeksProcarbazine(0,37.5,75or150mg/kg/day,treatmentforthree consecutivedays,oralgavage,analysesatdays15and30,the protocolincludedimmunodepletionofwild-typecells)

Dose-relatedincreasesinRETandRBCMfs.Atday15,theRETandRBCmeanMfs were~239106and~59106atthehighestdose[44] C57BL/6Jgpt deltamales, 6–8weeks

EMS(0,5,13,20,55and100mg/kg/day,28days,oralgavage; analysesoccurredatday29,onlyRBCswereanalysed)Dose-relatedincreaseinRBCMfsandasignificantincreaseinmeanRETMf.The RETmeanMfswere~209106 foralldoses≥13mg/kg/day,whereastheRBC meanMfsincreasedwithdoseandwere59106 atthehighestdose.Vehiclecontrol meanMfsweremaximum29106

[32] C57BL/6Ogg1+/+ andOgg1/ males, 10–11weeks

BaP(positivecontrolENU)(dosedwith50mg/kgbwfor3days, i.p.;positivecontrolENUwasgivenat22mg/kgbwfor3days, i.p.;analysed5dayspriortodosingandatdays16and34afterthe lastdosing,theprotocolincludedimmunodepletionofwild-type cellsandcomparisonswithpre-dosingmeasurementswerepossible)

SmallbutstatisticallysignificantlyincreasedRETandRBCMfs,mostevidentatday 16withnocleardifferencesbetweengenotypes[34] Atday16,themeanMfsofRETswere17.43and13.649106andRBCswere1.14 and1.259106fromOgg1+/+andOgg1/,respectively.Day5priortoexposure, themeanMfswere0.29and0.789106inRETsand0.13and0.089106in RBCsfromOgg1+/+andOgg1/,respectively.VehiclecontrolmeanMfsatday16 were7.70and0.279106inRETsand1.26and0.099106inRBCsfrom Ogg1+/+andOgg1/,respectively.MfsinENU-positivecontrolOgg1+/+miceatday 16were125.809106(RETs)and13.399106(RBCs) B6C3F1males, 7weeksAgNP(positivecontrolENU)NoincreasedMfinRETsorRBCs,despiteavailabilitytothebonemarrowandbone marrowcytotoxicity.Overthestudyperiod,RETandRBCmeanMfsinvehicle controlanimalsrangedfrom0to1.39106and0to0.49106.Nosignificant increasewasobservedintreatedanimals.PositivecontrolanimalstreatedwithENU hadmeanRETandRBCMfsat2weeksof5509106and1909106, respectively [39] (i.v.;singledosesof0.5,1.0,2.5,5.0,10.0or20.0mg/kg;positive controlENUwasgivenasasingledoseof140mg/kgbw.i.p.; analysesoccurredonday2andonweeks2,4and6afterdosing) C57BL/6Jp53+/+ (WT)andp53/ (KO)males1,7weeks

X-rays(160kVpwith0.5-mmCuand0.5-mmAlfilters)(1Gy, 0.52Gy/min,analysesoccurredatdays0,2,7,14,21,28,35,42, 49,56,63,77,91,112,133,225,323,413and713(onlyWTfor thelast4time-pointsduetodeathofKO);onlyRBCswere measured) IncreasedRBCMfinWTmiceandtwotimeshigherincreaseinKOmice,which declinedtobackgroundlevelsinWTmice.Themaximumresponseoccurredatday 28withRBCmeanMfsinWTandKOmiceof10.379106 and21.599106 , respectively.TheRBCmeanMfsofunexposedWTandKOatday28were 2.269106 and2.659106 ,respectively.Thereweremarkedinterindividual differencesinresponse,bothinthemagnitudeofthemaximalresponseandintime untilmaximumresponsewasevident

[36] (continued)

PIG-AIN MICE AND HUMAN CELLS 81

(5)

Table1.(continued) Strain(s),genotype(s), sex,ageAgent/intervention(exposure,route,assessmenttime).Methodological detailsMainoutcome,mutationfrequencies(Mf)inreticulocytes(RETs)orredbloodcells (RBCs)References B6C3F1males, 8weeks1,2-DCPandDCM(positivecontrolENU)(repeatedinhalation, B6C3F1:150,300and600ppm1,2-DCPor400,800and 1,600ppmDCM.Coexposuregroupswereexposedto 150+400ppmand300+800ppmof1,2-DCP+CMfor6hr/ day,fiveconsecutivedays/week,for6weeks;analysesoccurredin weeks3and6afterinhalationof1,2-DCPand/orDCM,onlyRBCs wereanalysed;positivecontrolENUwasgivenasasingledoseof 70mg/kgbwi.p.) NosignificantincreaseinRBCmeanMFforanytreatment.Positivecontrolanimals treatedwithENUhadRBCmeanMfsat222819106at3weeksand 2653129106at6weeks

[41] C57BL/6Ogg1+/ and Ogg1/ males, 8–11weeks

Selenium(Se)-deficiencythroughdiet.(positivecontrolENU)(Se depletionthroughtwogenerations.Miceweregivenadietofeither low(0.01mgSe/kgdiet)ornormal(0.23mgSe/kgdiet)Se content.PositivecontrolENUwasgiven22mg/kgbwfor3days, i.p.;theprotocolincludedimmunodepletionofwild-typecells.) SignificantlyincreasedRBCmenMf(p=0.008),butnotRETmeanMfinmicegiven thelowSediet.Nointeractionwasobservedbetweendietandgenotype.TheRET meanMfwasinsignificantlyincreasedinOgg1/ mice.ThemeanRETandRBC Mfswere179.99106and33.49106,respectively.

[42] Balb/cmales, 3weekFolate-deficient(D)/folate-supplemented(S)diet.(positivecontrol ENU)Micewerefedafolate-deficient(0mg/kg,D),control (2mg/kg,C)orsupplemented(6mg/kg,S)dietfromweaningfor 18weeks.PositivecontrolENUmicewasgiven80mg/kgbw,i.p.; theprotocolincludedimmunodepletionofwild-typecells)

NosignificantchangeinRBCorRETmeanMfsforDorScomparedtocontrol. IncreasedRBCmeanMfandnotRETMfinDcomparedtoS.Dmicehad2.2times higherRBCmeanMfthanC,althoughnotstatisticallysignificant.TheRBCmeanMf was3.8timeshigherinDmicecomparedtoS-mice(p=0.011).

[43] CD-1males, 8weeksENU,BaP,EC,P,MC(ENU:0,12.5,25,50mg/kg/day;BaP:62.5, 125,250mg/kg/day;EC:100,200,400mg/kg/day;P:125,250, 500mg/kg/day;MC:500,1000,2000mg/kg/day;givenonthree consecutivedaysbyoralgavage;analysesoccurredondays15and 30;theprotocolincludedimmunodepletionofwild-typecells)

Dose-relatedsignificantlyincreasedRETandRBCmeanMfsforthethreeknown mutagens(ENU,BaPandEC),nochangeforthetwoknownnon-mutagens(Pand MC)

[31] C57BL/6Jgpt deltamales,8weeksTiO2-NPs(positivecontrolENU)(i.v.;2,10or50mg/kgbw/week for4weeks;positivecontrolENUwasgivenasasingledoseof 70mg/kgbw.i.p.;analysesoccurredonday30afterfirstinjection, RBCswereassayed)

NoincreasedMfinRBCs,despiteavailabilitytothebonemarrowandbonemarrow cytotoxicity.RBCmeanMfinvehiclecontrolanimalswas0.400.559106 . PositivecontrolanimalstreatedwithENU(70mg/kg)hadmeanRBCMfof 511.49106

[40] B6C3F1males, 8weeksAcrylamide(positivecontrolENU)(givenat0,0.5,1.5,3.0,6.0, 12.0and24.0mg/kg/dayindrinkingwaterfor30days;positive controlENUwasgivenat10mL/kgviaoralgavageondays1–3 and29–30,andassayedonday31,;theprotocolincluded immunodepletionofwild-typecells.)

NoeffectdetectableinRETorRBCmeanMfs.ThepositivecontrolENUgaveriseto significantlyincreasedRETandRBCmeanMfs[45] (continued)

82 ANN-KARIN OLSENET AL.