Biotechnological uses of RNAi in plants: Risk assessment considerations

Biotechnological

uses

of

RNAi

in

plants:

risk

assessment

considerations

Josep

M.

Casacuberta

,

Yann

Devos

,

Patrick

du

Jardin

,

Matthew

Ramon

,

Herve´ Vaucheret

,

and

Fabien

Nogue´

1_{CenterforResearchinAgriculturalGenomics,ConsejoSuperiordeInvestigacionesCientı´ficas-InstitutdeRecercaiTecnologia}

Agroalimenta`ries-UniversitatAuto`nomadeBarcelona-UniversitatdeBarcelona,CampusUniversitatAuto`nomadeBarcelona, Bellaterra-CerdanyoladelValle`s,08193,Barcelona,Spain

2_{EuropeanFoodSafetyAuthority,ViaCarloMagno1/A,43126Parma,Italy}

3_{GemblouxAgro-BioTech,PlantBiologyUnit,UniversityofLie`ge,Gembloux,Belgium} 4

INRA,InstitutJean-PierreBourgin,SaclayPlantSciences,Versailles,France

5_{AgroParisTech,InstitutJean-PierreBourgin,SaclayPlantSciences,Versailles,France}

RNAi offers opportunities to generate new traits in

geneticallymodified(GM)plants.Insteadofexpressing

novel proteins, RNAi-based GM plants reduce target

gene expression. Silencing of off-target genes may

triggerunintended effects,and identifyingthesegenes wouldfacilitateriskassessment.However,using bioin-formaticsaloneisnotreliable,duetothelackofgenomic dataandinsufficientknowledgeofmechanisms

govern-ingmRNA–small(s)RNAinteractions.

Background

Technologicaladvancesingeneticsandgenomicsledtothe developmentofnewtypesGMplantinwhichtargetgenes aresilenced through RNAi. RNAi isaconserved mecha-nismthatreliesontheproductionofsRNA(20–30 nucleo-tidesinsize),whichpromotesdegradation ortranslation

repressionofhomologousmRNA.Inthecommon

eukary-oticancestor,RNAiwasprobablymeanttotarget molecu-lar parasites, such as viruses or transposons [1], but subsequently evolved to act on various aspects of gene regulationorgenomestructure[2].RNAicanoccur natu-rally, as a result of natural modifications of gene copy

number or genetic arrangements. Plant breeders have

already selected for natural RNAi-mediatedtraits, such asseedcoatcolourorlowglutelinlevel,insoybeanandrice, respectively [3], through conventional breeding without theaidofanytransgenicmanipulation.

However,transgenic RNAihasemergedas apowerful

andinnovativegenetictoolforbiological research.Ithas beenutilisedinfundamentalresearchfortheassessment ofgenefunctions,andinvariousfieldsofappliedresearch, suchas human and/or veterinarymedicine, and agricul-ture.Givenitspotentialtocontrolgeneexpression,RNAi offers newopportunities for plant breeding(Box 1), and isbeingused toalter agronomicandcompositional char-acteristics in GM plants (e.g., nutritional enrichment,

antinutrient reduction, and disease or pest resistance) [4].RNAi-basedGMplantsareapproaching commerciali-sation and, similar to GM plants in most jurisdictions, will likelybe subjecttoarisk analysisbeforeregulatory approval. Several international risk assessment bodies involvedinthepre-marketriskassessmentofGMplants arecurrentlyconsideringwhetherexistingapproachesare

appropriate or need to be refined for RNAi-based GM

plants(Table1).Here,wediscusspossibleriskstohuman andanimal healthandthe environment associatedwith the silencing of off-target genes, andpossible strategies toassesstheserisks.

RiskassessmentconsiderationsforRNAi-basedGM

plants

RisksassociatedwiththeintendedchangesincurrentGM plantsareusuallyrelatedtothenewlyexpressedproteins (e.g.,possibletoxicityandallergenicity).Giventhatmost RNAistrategiesaimtoreducetheamountofendogenous proteins,suchrisksarenotnecessarilyrelevantfor RNAi-based GM plants. However, the decreased expression of atargetgene mayhavesafety implicationsin particular cases(e.g.,ifasubstrateofasilencedenzymeaccumulates totoxiclevels).

RisksmorespecificallyassociatedwithRNAistrategies exist. Given that RNAi activity is based on the pairing

betweensRNA andmRNA,notonlythetargetgene,but

also off-target genes that produce mRNA with sufficient nucleotidesequencecomplementaritytothesRNAcouldbe silenced [5]. Owing tothe conserved natureofRNAi [6], targetand off-target genescould be silenced inboth the

RNAi-based GM plant and organisms consuming and/or

infectingthatplant.Theseorganismsincludetargetpests, inthe caseofhost-deliveredRNAi, andnontarget organ-isms(NTOs).Silencingoftargetandoff-targetgenesinthe GM plantandNTOscould lead toadverse effects,which require assessment. However, the likelihood ofsilencing agivengenewilldependontheabilityofthesRNAtobe asubstratefortheRNAimachineryoftherecipient organ-ism.Indeed,RNAimachineryvariesamongspecies[7]and thereisconvincingevidenceforonlyafeworganisms(e.g.,

nematodes and insects) that the plant-produced sRNA

uses host RNAi machinery [8]. It should be noted that

Science

&

Society

0167-7799/

ß2014ElsevierLtd.Allrightsreserved.http://dx.doi.org/10.1016/j.tibtech.2014.12.003

Correspondingauthor:Nogue´,F. (fabien.nogue@versailles.inra.fr).

Keywords:food/feedsafety;genesilencing;geneticallymodifiedorganism(GMO); nontargetorganism(NTO);off-targetgene;plantbreeding..

not alloff-target interactions will result in a significant reduction of in vivo gene expression. Even in the event that the expression levelof a gene is reduced, this may not necessarily result in detectable phenotypic changes. Only a fraction of such unexpected phenotypic changes maytranslateintoadverseeffects,dependingonthe func-tionofthesilencedoff-targetgene.

Given that silencing off-target genes within the GM plant or in organisms infecting or feeding on the plant may,insomecases,constituteanewriskforhumans,other animals,ortheenvironment,suchrisksshouldbeproperly assessedandmanaged.Genome-widebioinformatic anal-ysesofsequencecomplementaritybetweentheexpressed

sRNA(s) and all known mRNAs would help to identify

potential off-target genes in the GM plant itself and in

NTOs. However, there are several limitations to such

analyses.Despitethelargenumberofsequencedgenomes,

the genomes of many crop plants remain unknown. In

addition, for those genomes that have been sequenced, information is still limited, because variable fractions, usuallycomprisingcentromericorhighlyrepetitiveparts ofthegenome,remainunknown.

Formostsequencedspecies,onlyareferencegenomeis sequenced andassembledtosuchanextentthat itcould beconsidered relativelycomplete.Giventhat thesRNA– mRNAinteractionreliesonalimitedsequencelength,any sequence variability between varieties may change the spectrum ofpotentialtargetandoff-target genes.Owing to the lack of genome sequence information for all the varieties in whichthe RNAi-based transformation event may be introduced, it will not be possible to perform a complete searchofallpotentialoff-target genes.In addi-tion, because various species belonging todifferent taxa canbeexposedtoanRNAi-basedGMplant[9],thelimited availabilityofgenomedatafornon-modelspeciesfurther

reducestheusefulnessofabioinformatics-basedapproach forthe risk assessmentof NTOs.Owing tothese limita-tions, bioinformatic analyses are likely to be of limited valuetopredictoff-targetgeneeffectsintheGMplantorin NTOsexposedtothatplantinmostcases[10].However, bioinformaticscouldinformthespeciesselectionprocessby identifyinggroupsofNTOsthatarepotentiallysensitive. Acomplementaryapproachtobioinformaticsanalyses couldbetousenewmoleculartechniques,suchasparallel

analysis of RNA ends (PARE) [11]. PARE relies on the

ability of cleaved RNAs, such as those resulting from sRNA-guidedcleavageofamRNAbytheRNAimachinery, toligatetoRNAadaptorsbyvirtueoftheabsenceofacap attheir50_{end(cappedmRNAscannotbeligated).}

Sequenc-inglibrariesofcleavedmRNAsfromboththeRNAi-based

GM plant and an adequate comparator would allowthe

determinationoftheintended andunintendedtargets of

the sRNA(s) producedby the RNAi construct. However,

because the population of mRNAs depends on the cell

types, organs, developmental stages, or environmental conditions, a complete evaluation of possible off-target genes remains challenging. Moreover, this inventory of unintendedtargetsofthesRNA(s)producedbytheRNAi constructwouldberestrictedtotheGMplantitselfandnot likelyfeasibleinNTOs.

Concludingremarksandfutureperspectives

RNAiisanemergingtechnologythatoffersnew opportu-nities to plant breeders. The risk assessment of

RNAi-based GM plants presents some peculiarities compared

withthatofcurrentlycommercialisedGMplants. Allerge-nicityassessmentofnewlyexpressedproteins, for exam-ple,willnotbemeaningfulforRNAi-basedGMplants.By contrast, the evaluation of the potential risk associated with thesilencingofan off-target gene isspecific tothis

Box1.StrengthsandlimitationsofRNAiasatoolforplantbreeding

Strengths

TheefficacyofsilencingofatargetgeneviaRNAivariesfromone

GM plant toanother [13], which allows for theidentification of

plantsshowingarangeofdecreasedtargetgeneexpression.Thisis

particularlyinterestingwhencompletesilencingofthetargetgene

isnot viableorinduces anundesirable phenotype.Thismaybe

particularlyinterestingforthereductionofantinutrientlevelsinGM

plants.

The silencing ofa target genevia RNAi canbe spatially and/or

temporally controlled by suitable promoters, in contrast to a

conventionalgeneknockout.Itispossibletodownregulateagene

functioninparticulartissuesand/ororgans,whilepreservingitin

othertissuesand/ororgansordevelopmentalstages.

Manyimportantcropspecieshavegenomesoriginatingfromancestral

or recent polyploidisation events and, therefore, contain multiple

allelesofagene;RNAienablestheirsimultaneoussilencing[14].

TheRNAi processis,inpart, ubiquitousamong eukaryotes,and

double-strandedRNAorsmallinterferingRNA(siRNA)producedby

aplantcan,insomecases,silenceatargetgeneinanorganisms

feedingonorinfectingthatplant.Thisoffersnewopportunitiesfor

thecontrolofpestsandplantdiseases[15].

Limitations

Forreasonsthatarestillnotwellunderstood,somemRNAsappear

difficulttosilencethroughRNAi[13].

Given that rules governing the pairing of sRNAs with target

mRNAs are not fully known, off-target gene effects are

possible.

Besides triggering the degradation or translation repression of

homologousmRNA,RNAialsoinducesthemethylationof

homo-logous DNA,includingthetranscribedregion ofRNAitransgene

itself. DNA methylation sometimes spreads into adjacent

se-quencessuchasthetransgenepromoter,whichcouldeventually

reduceRNAiefficiencyovergenerations.

GiventhatRNAireliesonthesequence-specificpairingofsRNA

to mRNA,mutationsarising in thetarget mRNA in themiddle

of itscomplementary region tothe sRNA guide could prevent

targetingthismRNA,sometimeswithoutchangingthetranslated

amino acid sequence(silent mutations). Such mutationsact in

adominantmanner,whichincreasesthelikelihoodofoccurrence

in nature. This is of particular importance for RNAi-mediated

pest or disease control if RNAi relies on the production of a

single artificialmiRNA and not on a population ofsiRNA. For

example, mutations in the target sequence enable viruses to

circumventsilencingmediatedbyanartificialmiRNAproducedin

GMplants.

Mutations arising in an endogenous mRNA, close, but not

perfectly homologous to the target mRNA, could create a

complementary region. Having a novel target in GM plants

would dilute the effect of transgene-derived sRNA(s) on the

expected target.Such mutations wouldalso actin a dominant

manner.

Science

&

Society

technology. The effects of off-target gene silencing that significantlyimpactagronomicperformanceandcrop qual-ity canbedetected,andcounter-selected,bythe breeder duringthebreedingprocess,asisthecaseforthe develop-mentofanynewtraitinGMandconventionalplants,but other unintendedeffectscould persist inthe final RNAi-basedplant.Theidentificationofsuchpotentialoff-target effectscould beachievedby the identificationofmRNAs thatexhibitlimited,althoughsignificant,complementary tothesRNA molecules producedby theRNAi construct. Although new bioinformatics [12] and molecular techni-ques[11]mayhelptoevaluatethespectrumofoff-target sequences,theseanalysisarecurrentlyoflimitedvaluefor theriskassessmentforthreereasons:(i)suitablegenome dataareavailableonlyforalimitednumberofspecies;(ii) rulesgoverningefficientmRNA and/orsRNA recognition bytheRNAimachineryareincompletelyunderstood;and (iii)thecapacityofplantsRNAtotriggersilencingin non-plantorganismsisnotalwaysclearandhasbeenestimated foronlyafewspecies.ProgressinbasicresearchonRNAi mechanisms,productionofsuitablegenome datafor rele-vantspecies,anddesignofefficientalgorithmstomakemore reliable predictions will help to fill these technological gaps.This new informationand theassociated tools will notonly facilitatetheriskassessmentofRNAi-basedGM plants, butmay alsohavea moregeneral impact onthe evaluationofpotentialriskassociatedwiththeproduction ofsRNAinGMornon-GMplants.Despitethesetechnical limitationsandpossibleadvancements,theriskassessment strategies followed for current GM plants, based on the comparative analysis of the molecular, compositional, andagronomic/phenotypiccharacteristicsoftheGMplant and its conventional counterpart, remain applicable and adequatefortheevaluationofRNAi-basedGMplants.

Acknowledgements

TheauthorswouldliketothankElisabethWaigmannforcriticalreading

of themanuscript.F.N. wassupported bythe AgenceNational de la

Recherche(programANR-11-BTBR-0001-GENIUS).

Disclaimerstatement

Y.D.andM.R.areemployedbyEFSA.However,theviewsorpositions

expressedinthispublicationdonotnecessarilyrepresentinlegalterms

theofficialpositionofEFSA.EFSAassumesnoresponsibilityorliability

foranyerrorsorinaccuraciesthatmayappear.

References

1 Obbard,D.J.etal.(2009)TheevolutionofRNAiasadefenceagainst virusesandtransposableelements.Philos.Trans.R.Soc.Lond.B:Biol. Sci.364,99–115

2 Castel,S.E.and Martienssen,R.A. (2013)RNAinterference inthe nucleus:rolesforsmallRNAsintranscription,epigeneticsandbeyond. Nat.Rev.Genet.14,100–112

3 Erdmann, V.A. and Barciszewski, J. (2011) Non-Coding RNAs in Plants,Springer

4 Katoch,R. and Thakur,N. (2013) Advances in RNA interference technologyanditsimpactonnutritionalimprovement,diseaseand insectcontrolinplants.Appl.Biochem.Biotechnol.169,1579–1605

5 Senthil-Kumar,M.andMysore,K.S.(2011)CaveatofRNAiinplants: theoff-targeteffect.MethodsMol.Biol.744,13–25

6 Knip,M.etal.(2014)Trans-kingdomcross-talk:smallRNAsonthe move.PLoSGenet.10,e1004602

7 Saumet,A.andLecellier,C.H.(2006)Anti-viralRNAsilencing:dowe looklikeplants?Retrovirology3,3

8 Dua,C-G.etal.(2012)ApplicationofRNAsilencingtoplantdisease resistance.Silence3,5

9 Romeis,J. etal. (2014)Potential useof anarthropod database to supportthenon-targetriskassessmentandmonitoringoftransgenic plants.TransgenicRes.23,995–1013

10 Lundgren,J.G.andDuan,J.J.(2013)RNAi-basedinsecticidalcrops. Bioscience63,657–665

11 German,M.A.etal.(2008)GlobalidentificationofmicroRNA-target RNA pairs by parallelanalysis of RNA ends. Nat. Biotechnol.26, 941–946

12 Zhong,R.etal.(2014)Computationaldetectionandsuppression of sequence-specific off-target phenotypes from whole genome RNAi screens.NucleicAcidsRes.42,8214–8222

13 Kerschen, A. et al. (2004) Effectiveness of RNA interference in transgenicplants.FEBSLett.566,223–228

14 Rathore,K.S.etal.(2012)Ultra-lowgossypolcottonseed:generational stability of the seed-specific, RNAi-mediated phenotype and resumption of terpenoid profile following seed germination. Plant Biotechnol.J.10,174–183

15 Runo, S. (2011) Engineering host-derived resistance against plant parasites through RNA interference: challenges and opportunities. Bioeng.Bugs2,208–213

Table1.RiskassessmentconsiderationsonRNAi-basedGMplantsandderivedfoodand/orfeedproductsfrominternational bodies

Authorityand/orinstitute Webaddress

CenterforEnvironmentalRiskAssessment(CERA) http://cera-gmc.org/docs/cera_publications/pub_08_2011.pdf

EuropeanFoodSafetyAuthority(EFSA) http://www.efsa.europa.eu/fr/events/event/140604.htm

FoodStandardsAustraliaNewZealand(FSANZ) http://www.foodstandards.govt.nz/consumer/gmfood/Documents/

Heinemann%20Response%20210513.pdf

USEnvironmentalProtectionAgency(USEPA) http://www.epa.gov/scipoly/sap/meetings/2014/january/012814minutes.pdf

Science

&

Society